Fib Thesis

The Fibonacci Sequence Under Various Moduli
Marc Renault
May, 1996
A thesis submitted to
Wake Forest University
in partial fulfillment of the degree of
Master of Arts in Mathematics
Contents
1 Leonardo of Pisa and the Fibonacci Sequence
2 The Binet Formula
3
11
3 Modular Representations of Fibonacci

3.1 The Period . . . . . . . . . . . . . . .
3.2 The Distribution Of Residues . . . . .
3.3 The Zeros Of F(mod m) . . . . . . . .
Sequences
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
17
17
30
34
4 Personal Findings
47
4.1 Spirolaterals And The Fibonacci Sequence . . . . . . . . . . . . . . . . . . . . 47
4.2 Fibonacci Subsequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
A The First 30 Fibonacci and Lucas Numbers
63
B k(m), (m), and (m) for 2 m 1000
65
C One Period Of F (mod m) for 2 m 50
75
CONTENTS
Chapter 1
Leonardo of Pisa and the

Fibonacci Sequence
Fibonacci, the pen name of Leonardo of Pisa which means son of Bonacci, was born in Pisa,
Italy around 1170. Around 1192 his father, Guillielmo Bonacci, became director of the Pisan
trading colony in Bugia, Algeria, and some time thereafter they traveled together to Bugia.
From there Fibonacci traveled throughout Egypt, Syria, Greece, Sicily, and Provence where
he became familiar with Hindu-Arabic numerals which at that time had not been introduced
into Europe.
He returned to Pisa around 1200 and produced Liber Abaci in 1202. In it he presents
some of the arithmetic and algebra he encountered in his travels, and he introduces the placevalued decimal system and Arabic numerals. Fibonacci continued to write mathematical
works at least through 1228, and he gained a reputation as a great mathematician. Not
much is known of his life after 1228, but it is commonly held that he died some time after
1240, presumably in Italy.
Despite his many contributions to mathematics, Fibonacci is today remembered for the
sequence which comes from a problem he poses in Liber Abaci. The following is a paraphrase:
A man puts one pair of rabbits in a certain place entirely surrounded by a wall.
The nature of these rabbits is such that every month each pair bears a new pair
which from the end of their second month on becomes productive. How many
pairs of rabbits will there be at the end of one year?
If we assume that the first pair is not productive until the end of the second month, then
3
CHAPTER 1. LEONARDO OF PISA AND THE FIBONACCI SEQUENCE
clearly for the first two months there will be only one pair. At the start of the third month
the first pair will beget a pair giving us a total of two pair. During the fourth month the
original pair begets again but the second pair does not, giving us three pair, and so on.
Assuming none of the rabbits die we can develop a recurrence relation. Let there be Fn
pairs of rabbits in month n, and Fn+1 pairs of rabbits in month n + 1. During month n + 2,
all the pairs of rabbits from month n + 1 will still be there, and of those rabbits the ones
which existed during the nth month will give birth. Hence Fn+2 = Fn+1 + Fn . The sequence
which ensues when F1 = F2 = 1 is called the Fibonacci sequence and the numbers in the
sequence are the Fibonacci numbers.
n:
Fn :
1
1
2
1
3
2
4
3
5
5
6
8
7
13
8
21
9
34
10
55
11
89
12
144
...
...
Thus the answer to Fibonaccis problem is 144.

Interestingly, it was not until 1634 that this recurrence relation was written down by
Albert Girard.
Despite its simple appearance the Fibonacci sequence contains a wealth of subtle and
fascinating properties. For example,
Theorem 1.1 Successive terms of the Fibonacci sequence are relatively prime.
Proof: Suppose that Fn and Fn+1 are both divisible by a positive integer d.
Then their difference Fn+1 Fn = Fn1 will also be divisible by d. Continuing,
we see that d|Fn2 , d|Fn3 , and so on. Eventually, we must have d|F1 . Since
F1 = 1 clearly d = 1. Since the only positive integer which divides successive
terms of the Fibonacci sequence is 1, our theorem is proved.
One of the purposes of this chapter and the next is to develop many of the identities
needed in chapters three and four. All of these can be found in either [4] or [20]. Given
the recursive nature of the sequence, proof by induction is often a useful tool in proving
identities and theorems involving the Fibonacci numbers. One of the most useful is the
following.
Identity 1.2 Fm+n = Fm1 Fn + Fm Fn+1 .
Proof: Let m be fixed and we will proceed by inducting on n. When n = 1,

then Fm+1 = Fm1 F1 + Fm F2 = Fm1 + Fm which is true.
Now let us assume the identity is true for n = 1, 2, 3, . . . , k, and we will show
that it holds for n = k + 1. By assumption
Fm+k = Fm1 Fk + Fm Fk+1
and
Fm+(k1) = Fm1 Fk1 + Fm Fk .
Adding the two we get Fm+k + Fm+(k1) = Fm1 (Fk + Fk1 ) + Fm (Fk+1 + Fk )
which implies Fm+(k+1) = Fm1 Fk+1 + Fm Fk+2 which is precisely our identity
when n = k + 1.
As an example of this identity, we see that F12 = F8+4 = F7 F4 + F8 F5 = 13(3) + 21(5) =
144.
It is often useful to extend the Fibonacci sequence backward with negative subscripts.
The Fibonacci recurrence can be written as Fn = Fn+2 Fn+1 which allows us to do this.
n:
Fn :
...
...
-7
13
-6
-8
-5
5
-4
-3
-3
2
-2
-1
-1
1
0
0
1
1
2
1
3
2
4
3
...
...
Sequences such as the Fibonacci sequence which can be extended infinitely in both directions
are called bilateral.
With some inspection another useful identity presents itself:
Identity 1.3 Fn = (1)n+1 Fn .
Now we can combine the above two identities to obtain
Identity 1.4 Fmn = (1)n (Fm Fn+1 Fm+1 Fn )
Another important fact about the Fibonacci sequence is easily tackled with induction.
Theorem 1.5 Fm |Fmn for all integers m, n.
Proof: Let m be fixed and we will induct on n. If either m or n equals zero,
then the theorem is true by easy inspection. For n = 1 it is clear that Fm |Fm .
Now let us assume that the theorem holds for n = 1, 2, . . . , k and we will
show that it also holds for n = k + 1. Using identity 1.2 we see Fm(k+1) =
Fmk1 Fm + Fmk Fm+1 . By assumption Fm |Fmk , and so Fm divides the entire
right side of the equation. Hence Fm divides Fm(k+1) and the theorem is proved
for n 1. Since Fmn differs from Fmn by at most a factor of -1, then Fm |Fmn
for n 1 as well.
A surprising result, with a surprisingly simple geometric proof is demonstrated in the
following identity.
Identity 1.6 F12 + F22 + F32 + + Fn2 = Fn Fn+1 .
Proof: We can think of the squares of Fibonacci numbers as areas, and then
put them together in the manner below.
11 11
33
22
88
. . .
55
..
.
We can find the area of the above rectangle by summing the squares, F12 + F22 +
F32 + F42 + F52 + F62 , or by multiplying height times width, F6 (F5 + F6 ) = F6 F7 .
The general case for the sum of the squares of n Fibonacci numbers follows easily.
The following identity will be useful to us and it, too, can be proved geometrically.
2
Identity 1.7 Fn2 + Fn+1
= F2n+1 .
Proof:
Fn +Fn+1 =Fn+2
}|
Fn Fn
Fn+1 Fn+1
Fn+1 Fn = Fn1
The area above can be represented as Fn1 Fn+1 + Fn Fn+2 . From identity 1.2
this simplifies to Fn+(n+1) = F2n+1 .
Here is another identity involving the square of Fibonacci numbers.
Identity 1.8 Fn+1 Fn1 Fn2 = (1)n .
Proof:
Fn+1 Fn1 Fn2
= (Fn1 + Fn )Fn1 Fn2

2
= Fn1
+ Fn (Fn1 Fn )
2
= Fn1
Fn Fn2
2
= (Fn Fn2 Fn1
).
We can now repeat the above process on the last line to attain
2
(Fn Fn2 Fn1
)
2
= (1)2 (Fn1 Fn3 Fn2
)
2
(1)3 (Fn2 Fn4 Fn3
)
..
.
(1)n (F1 F1 F02 )
(1)n
It was the 19th century number theorist Edouard Lucas who first attached Fibonaccis
name to the sequence we have been studying. He also investigated generalizations of the
sequence.
A generalized Fibonacci sequence, G, is one in which the usual recurrence relation

Gn+2 = Gn+1 +Gn holds, but G0 and G1 may take on arbitrary values. The Lucas sequence,
L, is an example of a generalized Fibonacci sequence where L0 = 2 and L1 = 1. It continues
2, 1, 3, 4, 7, 11, .... There are many interesting relationships between the Fibonacci and
Lucas sequences, and we give two of the most basic here.
Identity 1.9 Ln = Fn1 + Fn+1 .
Proof: We will prove the identity by induction. It is easy to see that L1 = 1 =

0 + 1 = F0 + F2 and L2 = 3 = 1 + 2 = F1 + F3 . Now suppose that the identity
holds for Lr and Lr+1 :
Lr
Lr+1
= Fr1 + Fr+1
= Fr + Fr+2
Adding the two equations gives us Lr+2 = Fr+1 + Fr+3 and subtracting the top
equation from the bottom yields Lr1 = Fr2 + Fr . Thus the identity holds for
all positive and negative r.
Identity 1.10 F2n = Fn Ln .
Proof: F2n = Fn+n = Fn1 Fn + Fn Fn+1 = Fn (Fn1 + Fn+1 ) = Fn Ln .
We make a distinction between a Fibonacci sequence, meaning any generalized Fibonacci

sequence, and the Fibonacci sequence, meaning the sequence with G0 = 0 and G1 = 1.
Some authors generalize the sequence even more by using the relation Sn+2 = bSn+1 +
aSn . A further generalization examines sequences with the relation Sn = c1 Sn1 + c2 Sn2 +
+ ck Snk for constants k and ci . Throughout this paper we will concentrate primarily
on the Fibonacci sequence, though we will have occasion to make use of the generalized
form Gn+2 = Gn+1 + Gn . We end this chapter with two identities from [20] involving the
generalized Fibonacci sequence.
Identity 1.11 Gm+n = Fn1 Gm + Fn Gm+1 .

Proof: For the cases n = 0 and n = 1 we have
Gm
Gm+1
= F1 Gm + F0 Gm+1
= F0 Gm + F1 Gm+1
which is true since F1 = 1, F0 = 0, and F1 = 1. By adding the two equations

it is easy to see that our identity continues to hold for n = 2, 3, . . . and so on.
Subtracting the first equation from the second indicates that the identity also
holds for negative n.
Identity 1.12 Gmn = (1)n (Fn Gm+1 Fn+1 Gm )
Proof: This identity follows by substituting n for n in the above identity and
then using identity 1.3.
10
Chapter 2
The Binet Formula

While the recurrence relation and initial values determine every term in the Fibonacci
sequence, it would be nice to know a formula for Fn so we wouldnt have to compute all the
preceding Fibonacci numbers. Such a formula was discovered by Jacques-Philippe-Marie
Binet in 1843. Vajda [20] observes that this was actually a rediscovery since Abraham
DeMoivre knew about this formula as early as 1718. However, history has favored Binet
with the credit. Much of the material in this chapter can be found in [20].
Let us find the values for x which will give us the generalized Fibonacci sequence xn+2 =
xn+1 + xn . Since we are not concerned with the case where x = 0 which gives us the trivial
sequence, we may divide through by xn to attain x2 = x + 1, that is, x2 x 1 = 0. The
two roots of this equation are
=
1+ 5
2
1 5
.
2
(2.1)
Note the following properties of and :

+ =1
= 1
(2.2)
Now 1, , 2 , 3 , . . . and 1, , 2 , 3 , . . . are in fact generalized Fibonacci sequences since

n+2 = n+1 + n and n+2 = n+1 + n . Indeed, any linear combination of n and n
forms the nth term of some Fibonacci sequence.
Gn = n + n
(2.3)
As we see, ( n + n )+( n+1 + n+1 ) = ( n + n+1 )+( n + n+1 ) = n+2 + n+2 .

11
12
CHAPTER 2. THE BINET FORMULA
Any Fibonacci sequence can be expressed this way for particular values of and . We
show this by expressing and in terms of G0 , G1 , , and .
First, notice that G0 = + and G1 = + . Since = G0 we can write
G1
= + (G0 )
= ( ) + G0
= 5 + G0
which implies
=
G1 G0
.
5
(G0 ) +
(2.4)
Similarly = G0 and so
G1
= G0 + ( )
= G0 5
which implies
=
G0 G1
.
5
(2.5)
Now, once we know G0 and G1 we can use equations 2.4 and 2.5 to determine and , and
then the formula for Gn follows from equation 2.3.
For example, in the Fibonacci sequence F0 = 0 and F1 = 1, and so = 1/ 5 and
= 1/ 5. Thus
n n
Fn =
.
(2.6)
5
In the Lucas sequence L0 = 2 and L1 = 1.
=
( + ) 2

1 2
=
= = 1.
5
5
5
2 ( + )

2 1
=
= = 1.
5
5
5
Thus
Ln = n + n .
(2.7)
DeMoivre was able to derive the formula for Fn in a different way, using generating
functions. We demonstrate this technique next.
13
Let g(x) =
i=0
Fi xi . It follows that g(x) F0 x0 F1 x1 = g(x) x. Hence

g(x) x =
Fi xi =
i=2
= x
(Fi1 xi + Fi2 xi )
i=2
Fi xi + x2
i=1
Fi xi
i=0
= xg(x) + x2 g(x).
Now we have g(x) xg(x) x2 g(x) = x. That is,
g(x)
Expressing
1
1 x
and
g(x)
x
x
=
1 x x2
1 ( + )x + x2
( )x
x
=
(1 x)(1 x)
5(1 x)(1 x)
1 x
1 x
5(1 x)(1 x)
5(1 x)(1 x)
1
1
.
5(1 x)
5(1 x)
1
1x
as the sums of geometric series we get
1
1
(1 + x + 2 x2 + ) (1 + x + 2 x2 + )
5
5
[( )x + ( 2 2 )x2 + ]/ 5
The coefficient of xn , in other words Fn , is ( n n )/ 5, just as we suspected.

We can use the generating function to attain some unusual results. Taking our equation
X
x
g(x) =
=
Fi xi
1 x x2
i=0
and dividing through by x we get
X
1
=
Fi xi1 .
2
1xx
i=0
When x = 1/2,
4=
X
Fi
i1
2
i=0
(2.8)
14
which implies
2=
X
Fi
i=0
2i
(2.9)
Another remarkable summation identity is obtained by differentiating both sides of equation (2.8)
X
1 + 2x
=
(i 1)Fi xi2 .
(1 x x2 )2
i=0
When x = 1/2,
32
X
(i 1)Fi
i=0
2i2
X
(i 1)Fi
2i
i=0
X
iFi
i=0
10
2i
X
iFi
i=0
2i
X
iFi
i=0
2i
X
Fi
i=0
2i
(2.10)
The next identity is clearly more complicated than those weve looked at before, yet its
proof yields readily to the Binet formula. The interesting thing about it is that it gives us
insight into the recurrence relation governing subsequences of the Fibonacci sequence. This
identity shows how every nth term of the Fibonacci sequence is related.
15
Identity 2.1 Fm+n = Ln Fm + (1)n+1 Fmn .

Proof:
Ln Fm + (1)n+1 Fmn = ( n + n )(
mn mn
m m
)
) + (1)n+1 (
5
5
(1)n+1 mn (1)n+1 mn
m+n n m + n m m+n
+
5
5
m+n (1)n mn + (1)n mn m+n (1)n mn + (1)n mn
m+n m+n
= Fm+n
It is easy to see that when n = 1 in the above identity, we get the usual Fibonacci
recurrence relation, Fm+1 = (1)Fm + (1)Fm1 . When n = m we get identity 1.10: F2n =
Ln Fn .
Identity 2.2 Fn =
n
1
2n1 [ 1
n
3
5+
n
5
52 +
n
7
53 + ].
Proof:
1
n n
= [(1 + 5)n (1 5)n ]

5
2n 5
Now expand using the binomial theorem:
n 2 n 3

n
1 h
=
1+
5+
5 +
5 +
1
2
3
2n 5
Fn =

n
n 2 n 3
i
1
5+
5
5 +
2
3
1
=
h n
1
2n1 5
1
h n
2n1
5+
n
3
n
3
5+
n
5
5 +
n
52 +
5
n
7
i
5
5 +
i
53 +
Lastly in this chapter we use and to demonstrate some simple greatest-integer identities. Though we will not use these, much research has been done in this area and they are
certainly of interest in their own right.
16
+ 12 c for all n.
Identity 2.3 Fn = b
5
Proof: |Fn
|
5
|<
= |
5
1
2
for all n.
Identity 2.4 Fn+1 = b Fn + 12 c for n 2.

Proof: |Fn+1 Fn | = |
n 2.
n+1
n+1
n+1 n
|
5
= |
( )
|
5
= | n | <
1
2
for all
Chapter 3
Modular Representations of
Fibonacci Sequences
One way to learn some fascinating properties of the Fibonacci sequence is to consider the
sequence of least nonnegative residues of the Fibonacci numbers under some modulus. One
of the first modern inquiries into this area of research was made by D. D. Wall [22] in 1960,
though J. L. Lagrange made some observations on these types of sequences in the eighteenth
century. Typically, the variable m will be used only to denote a modulus.
3.1
The Period
Perhaps the first thing one notices when the Fibonacci sequence is reduced mod m is that
it is periodic. For example,
F (mod 4)
F (mod 5)
=
=
0 1 1 2 3 1 0 1 1 2 3 ...
0 1 1 2 3 0 3 3 1 4 0 4 4 3 2 0 2 2 4 1 0 1 1 2 3 ...
See appendix C for a list of the Fibonacci sequence under various moduli.
Any (generalized) Fibonacci sequence modulo m must repeat. After all, there are only
2
m possible pairs of residues and any pair will completely determine a sequence both forward
and backward. If we ignore the pair 0,0 which gives us the trivial sequence, then we know
that the period of any Fibonacci sequence mod m has a maximum length of m2 1.
It will always happen that the first pair to repeat will be the pair we started with.
Suppose that this were not so. Then we might have the sequence a, b, ..., x, y, ..., x, y, ...
where the pair a, b is not contained in the block x, y, ..., x, y. However, we know that this
17
18
CHAPTER 3. MODULAR REPRESENTATIONS OF FIBONACCI SEQUENCES
block repeats backward as well as forward, and so the pair a, b cannot be in the sequence.
This gives us our contradiction.
We can say some things about where the zeros will appear in the modular representation
of the Fibonacci sequence. Recall from identities 1.2 and 1.4 that
Fs+t
= Fs1 Ft + Fs Ft+1
Fst
= (1)t (Fs Ft+1 Fs+1 Ft ).
If Fs Ft 0 then clearly Fs+t 0 and Fst 0. Hence all the zeros of F (mod m) are
evenly spaced throughout the sequence. Since F (mod m) is periodic for any m and F0 = 0
we can say that any integer will divide infinitely many Fibonacci numbers. In addition,
all the Fibonacci numbers divisible by a given integer are evenly spaced throughout the
sequence.
We know that F (mod m) is periodic, so the question naturally presents itself: What is
the relationship between the modulus of a sequence and its period? We will examine some
results in this area.
Each author seems to have his or her own notation, but the following definitions come
from Wall. Let k(m) denote the period of the Fibonacci sequence modulo m. Let h(m)
denote the period of any generalized Fibonacci sequence modulo m. From our previous example we see that k(4) = 6 and k(5) = 20. The following are some immediate consequences
of the definition.
Fn
Fn+rk(m)
(mod m)
Gn
Gn+rh(m)
(mod m)
Fk(m)
Fk(m)1
0 (mod m)
Fk(m)+1 Fk(m)+2 1
(3.1)
(mod m)
(3.2)
We will often use the fact that if Fn 0 (mod m) and Fn+1 1 (mod m) then k(m)|n.
This result follows immediately from the periodicity of F (mod m).
We now demonstrate some very general properties of F (mod m) using our notation.
The following three theorems can be found in [22]. The reader is encouraged to examine the
table in appendix B which gives the period of F (mod m) for 2 m 1000, and observe the
3.1. THE PERIOD
19
behavior of k(m) as m varies. After noticing that F (mod m) is periodic one notices that
almost all of the periods are even. Though Wall provides the next theorem, the proof is the
authors.
Theorem 3.1 For m 3, k(m) is even.
Proof: For ease of notation let k = k(m), and we will consider all congruences
to be taken modulo m. From identity 1.3 we know that if t is odd then Ft = Ft
and if t is even then Ft = Ft . We will assume k to be odd and show that m
must equal 2.
We know that F1 = F1 Fk1 . Now k 1 is even so Fk1 = F1k F1 .
Thus F1 F1 , and as a result m = 2.
Another curious feature of F (mod m) is that k(n)|k(m) whenever n|m. Surprisingly,
this property is true of generalized Fibonacci sequences as well.
Theorem 3.2 If n|m, then for a given Fibonacci sequence, h(n)|h(m).
Proof:
Let h = h(m). We need to show that G(mod n) repeats in blocks
of length h. We do this by showing that Gi Gi+h (mod n) regardless of our

choice for i. Certainly we know that Gi Gi+h (mod m), so for some 0 a < m
we have Gi = a + mx and Gi+h = a + my.
Now say m = nr and let us substitute nr for m in the previous two equations.
Then Gi = a+nrx and Gi+h = a+nry. We can say that a = a0 +nw (0 a0 < n)
and this time substitute for a in the previous equations. Now Gi = a0 +n(w+rx)
and Gi+h = a0 +n(w+ry). Of course this implies that Gi Gi+h (mod n) which
was needed to be shown.
We can make this theorem more exact by expressing h(m) in terms of h(pei i ) where m
Q
has the prime factorization m = pei i .
Theorem 3.3 Let m have the prime factorization m =
the least common multiple of the h(pei i ).
pei i . Then h(m) = lcm[h(pei i )],
20
Proof: By our previous theorem h(pei i )|h(m) for all i. It follows that lcm[h(pei i )]|h(m).
Second, since h(pei i )|lcm[h(pei i )] we know G(mod pei i ) repeats in blocks of
length lcm[h(pei i )]. Hence Glcm[h(pei )] G0 and Glcm[h(pei )]+1 G1 (mod pei i )
i
for all i. Since all the
pei i
are relatively prime, the Chinese Remainder Theo-
rem assures us that Glcm[h(pei )] G0 and Glcm[h(pei )]+1 G1 (mod m). Thus
i
G(mod m) repeats in blocks of length lcm[h(pei i )] and we can say that h(m)|lcm[h(pei i )].
This concludes the proof.
Hence we have reduced the problem of characterizing k(m) into the problem of characterizing k(pe ). Before we develop theorems which speak to this problem, however, we look
at a related result. We see in the following theorem that it is not necessary to break a
modulus all the way into its prime factorization in order to attain information about k(m).
Theorem 3.4 h([m, n]) = [h(m), h(n)] where brackets denote the least common multiple
function.
Proof: Since m|[m, n] and n|[m, n] we know that h(m)|h([m, n]) and h(n)|h([m, n]).
It follows that [h(m), h(n)]|h([m, n]).
Say we have the prime factorization [m, n] = pe11 . . . pet t . Then h([m, n]) =
h(pe11 . . . pet t ) = [h(pe11 ) . . . h(pet t )]. Since pei i divides m or n for all i, certainly
h(pei i ) divides h(m) or h(n) for all i. Thus [h(pe11 ) . . . h(pet t )]|[h(m), h(n)]. In
other words, h([m, n])|[h(m), h(n)].
Hence h([m, n]) = [h(m), h(n)].
Now we turn our attention back to the matter of solving k(pe ) in terms of pe . The general
case of this theorem is somewhat involved, so to motivate the ideas we look at a couple of
specific cases. We will demonstrate that k(2e ) = 32e1 and k(5e ) = 45e . These results can
be found in [11], but the proofs there are somewhat incomplete. For example, Kramer and
Hoggatt show that F32e1 0 (mod 2e ) and F32e1 +1 1 (mod 2e ) and they immediately
conclude that k(2e ) = 3 2e1 . However, this only demonstrates that k(2e )|3 2e1 . The
proof below takes this point into consideration.
Theorem 3.5 k(2e ) = 3 2e1 .
3.1. THE PERIOD
21
Proof: By inspection, k(2) = 3 and k(4) = 6, so the theorem holds for e = 1, 2.

Suppose k(2r ) = 3 2r1 , so that F32r1 0 and F32r1 +1 1 (mod 2r ) and
we will induct on r.
By identity 1.10,
F32r = (F32r1 )(F32r1 1 + F32r1 +1 )
The first factor on the right, F32r1 0 (mod 2r ). It is an easy matter to see
that every third Fibonacci number is even and the rest are odd, so the second
factor on the right, F32r1 1 + F32r1 +1 0 (mod 2). Thus their product,
F32r 0 (mod 2r+1 ).
By identity 1.7,
F32r +1
(F32r1 )2
(0 or 2r )2
0
1
+
+
+
(F32r1 +1 )2
(1 or 2r + 1)2
1
(mod 2r+1 )
(mod 2r+1 )
(mod 2r+1 )
Thus we know k(2r+1 )|3 2r .

We know that k(2r )|k(2r+1 ), and by our induction hypothesis k(2r ) = 3
2r1 . Thus k(2r+1 ) = either 3 2r1 or 3 2r . We will show that the latter is
true by proving F32r1 +1 6 1 (mod 2r+1 ). More precisely, we will prove that
F32r1 +1 2r + 1 (mod 2r+1 ).
Let us add this statement to our induction hypothesis: assume F32r2 0
and F32r2 +1 2r1 + 1 (mod 2r ). By inspection this is the case for r = 3. We
will induct on r to show that F32r1 +1 2r + 1 (mod 2r+1 ).
By identity 1.7,
F32r1 +1 = (F32r2 +1 )2 + (F32r2 )2 .
Let us look at the first term on the right.
F32r2 +1
(2r1 + 1) or (2r1 + 1 + 2r )
(2r1 + 1)2
22r2 + 2r + 1 2r + 1
(2r1 + 1 + 2r )2
(3 2r1 + 1)2
(mod 2r+1 )
(mod 2r+1 )
= 9 22r2 + 3 2r + 1 2r + 1
(mod 2r+1 )
22
Thus
(F32r2 +1 )2 2r + 1 (mod 2r+1 ).
Let us look at the second term on the right.
F32r2
02
(2r )2
0 or 2r
(mod 2r+1 )
0 (mod 2r+1 )
=
22r 0 (mod 2r+1 )
Thus
(F32r2 )2 0 (mod 2r+1 ).
Consequently F32r1 +1 2r + 1 (mod 2r+1 ) and the theorem follows.
Before we prove a similar theorem for k(5e ) we first need to mention a lemma which
provides insight into one of the divisibility properties of the Fibonacci sequence.
Lemma 3.6 Let p be an odd prime and suppose pt |Fn but pt+16 | Fn for some t 1. If p6 | v
then pt+1 |Fnvp but pt+26 | Fnvp .
The proof of this lemma is too involved to be examined here, but it can be found in [20].
Since F5 = 5, clearly 5 goes into F5 once. By the the lemma then, 5 goes into Fn5t
exactly t times if 56 | n.
Theorem 3.7 k(5e ) = 4 5e .
Proof: First we will show that for all e, F45e 0 and F45e +1 1 (mod 5e ).
By the lemma above clearly F45e 0 (mod 5e ). From identity 1.7 we have
F45e +1 = (F25e )2 + (F25e +1 )2 .
Applying our lemma to the first term on the right gives us (F25e )2 02
0 (mod 5e ). By identity 1.8, the second term on the right, (F25e +1 )2 = F25e F25e +2 +
(1)25
+2
1 (mod 5e ).
Thus we know that F45e 0 and F45e +1 1 (mod 5e ) and it follows that
k(5e )|4 5e for all e.
3.1. THE PERIOD
23
We proceed now by induction using the hypothesis k(5r ) = 4 5r . By inspection this is the case for r = 1. We induct on r to show k(5r+1 ) = 4 5r+1 .
Since 4 5r = k(5r )|k(5r+1 ) and k(5r+1 )|4 5r+1 , we know k(5r+1 ) = either
4 5r or 4 5r+1 . In the first case, F45r contains exactly r factors of 5 and hence
F45r 6 0 (mod 5r+1 ). Hence we must have k(5r+1 ) = 4 5r+1 and our theorem
is proved.
We have proved k(pe ) = pe1 k(p) for p = 2 and p = 5, and we now turn to proving it
for all p. While the general theorem is slightly less strict than these special cases, it does
give us great insight into the periodic behavior of the Fibonacci sequence. One proof, by
Robinson[15], also shows how matrices may be used to discover facts about the Fibonacci
sequence. We will be working with the matrix U defined by

0 1
U=
1 1
which has the property that
n
U =
Fn1
Fn
Fn
Fn+1

.
(3.3)
This matrix has been called the Fibonacci matrix because of this property. Note that
U k(m) = I + mB I (mod m) for some 2 2 matrix B. And, if U n I (mod m) then
k(m)|n.
Theorem 3.8 If t is the largest integer such that k(pt ) = k(p) then k(pe ) = pet k(p) for
all e t.
e
Proof: Since U k(p ) = I + pe B we can write U pk(p ) = (I + pe B)p = I p +

p1 e

e
p
(p B) + p2 I p2 (pe B)2 + . Thus U pk(p ) I (mod pe+1 ). Clearly
1 I
e+1
U k(p
I (mod pe+1 ) and so we have k(pe+1 )|pk(pe ). Combine this fact with
the knowledge that k(pe )|k(pe+1 ) and as a result k(pe+1 ) = k(pe ) or pk(pe ).
Now let us assume that k(pe+1 ) = pk(pe ) and we will induct on e to arrive
at k(pe+2 ) = pk(pe+1 ).
e
By assumption k(pe ) 6= k(pe+1 ) and so we know that U k(p ) = I + pe B where

e
p6 | B. Hence U pk(p ) = (I + pe B)p = I p + p1 I p1 (pe B) + p2 I p2 (pe B)2 +
I + pe+1 B (mod pe+2 ). (The astute reader may notice that this congruence does
24
not actually hold for p = 2, e = 1. However, since we have proven the case for
e+1
p = 2 in theorem 3.5 we may assume p 3.) That is, U k(p
= U pk(p
I + pe+1 B 6 I (mod pe+2 ). This implies k(pe+1 ) 6= k(pe+2 ) so we must have

k(pe+2 ) = pk(pe+1 ) and the induction is complete.
Now let t be the largest e such that k(pe ) = k(p). Then k(pe ) = k(p) for
1 e t and k(pe ) = k(pet ) for e t.
Remarkably, the conjecture that t = 1 for all primes has existed since Walls paper
in 1960, but neither a proof nor a counter example has yet been found. Once again we
have been able to reduce the problem of finding the period of the Fibonacci sequence given
a modulus. All that remains (other than proving t = 1 for all primes, of course) is to
characterize k(p) in terms of p. However, this undertaking has proven to be extraordinarily
difficult, and the best we can do is to describe some bounds on k(p).
Theorems 3.11 through 3.13 will describe upper bounds on k(p), but we must first take
some time to develop certain divisibility properties of the Fibonacci sequence found in [20].
We will make use of these three facts for primes, p:
p
0 (mod p) for 1 n p 1
n

p1
(ii)
(1)n (mod p) for 0 n p 1
n

p+1
0 (mod p) for 2 n p 1
(iii)
n
(i)
We will also use the following lemma:

Lemma 3.9 5 is a quadratic residue modulo primes of the form 5t 1 and a quadratic
nonresidue modulo primes of the form 5t 2.
Proof:
Using the Legendre symbol and the law of quadratic reciprocity we
know that ( p5 ) = ( p5 ) since 5 is a prime of the form 4t + 1. Then,

5
5t + 1
5
5t 1
5t + 1
5
5t 1
5

1
=
=1
5

4
= 1.
5
3.1. THE PERIOD
25
However,

5
5t + 2

=
and Fermats theorem we get Fp 5

5
p1
2

=

2
= 1
5

3
=
=
= 1.
5
p1

Using identity 2.2 we have 2p1 Fp = p1 + p3 5 + + pp 5 2 . Applying (i) above
5
5t 2
5t + 2
5
p1
2
5t 2
5
(mod p). Now from the lemma we know that
1 (mod p) if and only if p = 5t 1, and 5
p1
2
1 (mod p) if and only if p = 5t 2.
Hence,
If a prime p is of the form 5t 1 then Fp 1 (mod p).
If a prime p is of the form 5t 2 then Fp 1 (mod p).
The converse of these statements is not true, for in fact F22 = 17711 = 85(22) + 1
1 (mod 22), and quite clearly, 22 is not prime.
Again we use identity 2.2 and see 2p2 Fp1 =
(ii) above, 2p2 Fp1 (1 + 5 + 52 + + 5
p3
2
p1
1
p1
3
5 + +
p1
p2
p3
2
. By
) (mod p). Summing the geometric series
yields,
2
Clearly if 5
p1
2
p2
Fp1
p1
2
(mod p).
1 (mod p) then 2p2 Fp1 0 (mod p). Certainly 2p2 6 0 (mod p) so it
would have to be that Fp1 0 (mod p). Hence,

If a prime p is of the form 5t 1 then Fp1 0 (mod p).

p1

Finally, identity 2.2 gives us 2p Fp+1 = p+1
+ p+1
5 + + p+1
5 2 . By (iii)
1
3
p

p1

p1
p1
above, 2p Fp+1 p+1
+ p+1
5 2 = (p + 1) + (p + 1)5 2 1 + 5 2 (mod p). Now
1
p
when 5
p1
2
1 (mod p) we will have 2p Fp+1 0 (mod p). Since 2p 2 (mod p), it would
have to be that Fp+1 0 (mod p). Thus,

If a prime p is of the form 5t 2 then Fp+1 0 (mod p).
Collecting our results provides us with the following theorem.
Theorem 3.10 If a prime p is of the form 5t 1 then Fp1 0 and Fp 1 (mod p). If a
prime p is of the form 5t 2 then Fp 1 and Fp+1 0 (mod p).
26
Now at last we can present some theorems concerning the character of k(p).
Theorem 3.11 If p is a prime, p 6= 5, then k(p)|p2 1.
Proof:
By inspection, k(2) = 3, and the theorem holds for p = 2. Now let
us assume p odd. To prove the theorem we will show that Fp2 1 0 and
Fp2 1 (mod p).
If p = 5t 1 then p|Fp1 . We know that p 1|p2 1 and so Fp1 |Fp2 1 .
Hence p|Fp2 1 . If p = 5t 2 then p|Fp+1 . We know that p + 1|p2 1 and so
Fp+1 |Fp2 1 . Hence p|Fp2 1 . Thus for all p 6= 5, Fp2 1 0 (mod p).
2 2
2
To prove Fp2 1 (mod p), we first note that 2p 1 Fp2 = p1 + p3 5 +
2 p2 1
2
2
+ pp2 5 2 . Now 2p 1 = (2p1 )p+1 1 (mod p) and pn 0 (mod p)
for 1 n p2 1. Hence Fp2 (5
p1
2
)p+1 (1)p+1 1 (mod p).
Thus the theorem is proved.
Theorem 3.12 If a prime p is of the form 5t 1 then k(p)|p 1.
Proof:
Since p = 5t 1, theorem 3.10 tells us that Fp1 0 and Fp
1 (mod p). Hence F (mod p) repeats in blocks of length p 1. Thus k(p)|p 1.
3.1. THE PERIOD
27
Theorem 3.13 If a prime p is of the form 5t 2 then k(p)|2p + 2.

Proof: Suppose p = 5t 2. By theorem 3.10 then, Fp+1 0 (mod p). Since
p + 1|2p + 2 and the zeros of F (mod p) are evenly spaced, we can say F2p+2
0 (mod p).
We also know that Fp 1 (mod p). By the usual recurrence relation
then, Fp+2 Fp+3 1 (mod p). Now we can write F2p+3 = F(p+1)+(p+2) =
Fp Fp+2 + Fp+1 Fp+3 (1)(1) + (0)(1) 1 (mod p).
Thus k(p)|2p + 2.
We now consider upper and lower bounds for k(m) where m can be any integer m 2.
It was noted earlier that k(m) m2 1, but can we do better than this? One method of
determining an upper bound is to use the two preceding theorems to show that k(p) = p2 1
if and only if p = 2 or 3. Then by application of theorem 3.8 it can be proved that k(pe ) <
(pe )2 1 for e 2. Finally, we can use theorem 3.4 to demonstrate that k(m) < m2 1 for
all composite m.
However, the author would like to propose a more elegant way of proving this inequality.
If k(m) = m2 1 for some m, then every possible pair of residues except 0, 0 appears in
F (mod m). It will be seen in section 3.3 that a single period of F (mod m) contains at most
four zeros. Consequently, F (mod m) contains at most four 0, r pairs where 1 r < m.
Hence, for m 6 there is some 0, r pair not represented in F (mod m) and we must have
k(m) < m2 1. It is a simple matter to check the remaining cases. When m = 2 or 3,
k(m) = m2 1 and when m = 4 or 5, k(m) < m2 1.
Unfortunately, empirical evidence indicates that this upper bound becomes less and less
precise as m increases. For 2 m 299 we find by inspection that k(m) 6m. For
300 m 1000 we find k(m) 4m. There does not appear to be any other material
published concerning upper bounds on k(m).
A sharp lower bound on k(m) was given by Paul Catlin[5] in 1974. We see here an
interesting interaction between the Lucas and Fibonacci numbers. We present his theorem
without proof.
28
Theorem 3.14 Given a modulus m > 2, let t be any natural number such that Lt m.
Then k(m) 2t with equality if and only if Lt = m and t is odd.
In the beginning of the section we saw some results which helped to define the overall
character of h(m) and thus of k(m). In the last part of this section we present some results
due to Wall on the relationship between h(m) and k(m). The first of these is strikingly
simple.
Theorem 3.15 h(m)|k(m).
Proof: For ease of notation, let k = k(m). Applying identity 1.11 and then
equations 3.1 and 3.2 we get Gk = Fk1 G0 + Fk G1 G0 (mod m) and Gk+1 =
Fk G0 + Fk+1 G1 G1 (mod m). Thus G(mod m) repeats in blocks of length
k(m). From this we can conclude h(m)|k(m).
We can elicit equality between h(m) and k(m) by putting stricter conditions on m.
Theorem 3.16 If a prime p is of the form 5t 2 then h(pe ) = k(pe ).
Proof: For ease of notation, let h = h(pe ). Certainly,
Gh G0
(mod pe ), and
Gh1 G1
(mod pe ).
By identity 1.11 the above is equivalent to

(Fh1 G0 + Fh G1 ) G0
= G1 Fh + G0 (Fh1 1) 0
(mod p)
(Fh1 G1 + Fh+1 G2 ) G1
= G2 Fh + G1 (Fh1 1) 0
(mod p)
We now have a system of equations which we may treat as a matrix with

determinant D = G21 G0 G2 . We will assume that D 0 (mod p) and arrive
at a contradiction.
When D 0 (mod p) then G21 G0 G2 = G21 G0 (G0 + G1 ) = G21 G0 G1
G20 0 (mod p). This last congruence is impossible if p = 2 (we ignore the
3.1. THE PERIOD
29
trivial case G0 = G1 = 0) so we may assume p to be odd. Multiply both sides

by 4.
4G20 + 4G0 G1 4G21 0
(mod p)
Add 5G21 to both sides.

4G20 + 4G0 G1 + G21 = (2G0 + G1 )2 5G21
(mod p)
and apply Fermats theorem to the right.

Now multiply both sides by Gp3
1
(2G0 + G1 )2 5
Gp3
1
(mod p)
Since p is odd, this implies that 5 is a quadratic residue modulo p. However,

by lemma 3.9, this contradicts our hypothesis that p = 5t 2. Therefore,
D 6 0 (mod p), and this implies that D 6 0 (mod pe ).
Since the determinant of our matrix is not congruent to zero (mod pe ), it
must have a unique solution modulo pe . Certainly Fh 0 and (Fh1 1)
0 (mod pe ) satisfy the system, and so Fh 0 and Fh1 1 (mod pe ). From the
recurrence relation, Fh+1 1 (mod pe ) as well.
Since h = h(pe ) our results tell us that k(pe )|h(pe ). By theorem 3.15 we
know h(pe )|k(pe ). Hence h(pe ) = k(pe ).
Wall proved several other results regarding the relationship between h(m) and k(m). We
present some of these here without proof. As usual, p denotes a prime number.
Theorem 3.17 Let D = G21 G0 G1 G20 . If gcd(D, m) = 1 then h(m) = k(m).
Theorem 3.18 If h(pe ) = 2t + 1 for some generalized Fibonacci sequence, then k(pe ) =
4t + 2.
Theorem 3.19 If p > 2 and k(pe ) = 4t + 2 then there exists some generalized Fibonacci
sequence where h(pe ) = 2t + 1.
Theorem 3.20 Let p be odd and p 6= 5. If h(pe ) is even then h(pe ) = k(pe ).
We end the section with two curious theorems pertaining to the period of F (mod m)
which did not seem to fit logically elsewhere within this section. We defer their proofs to the
30
end of section 3.3 where we will have developed other ideas sufficiently to make their proofs
easy. The curious feature here is that the modulus of the sequence is an actual Fibonacci
or Lucas number. Theorem 3.45 can be found in [7] and theorem 3.46 can be found in [17].
Theorem 3.45 If n 5 is odd then k(Fn ) = 4n. If n 4 is even then k(Fn ) = 2n.
Theorem 3.46 If n 3 is odd then k(Ln ) = 2n. If n 2 is even then k(Ln ) = 4n.
These theorems indicate that for any even t 6 there is some m such that k(m) = t.
Also, k(m) can not equal 4, otherwise F4 = 3 0 and F3 = 2 1 (mod m) which is not
possible. Hence the range of k(m) is 3 union the set of all even numbers greater than or
equal to 6.
3.2
The Distribution Of Residues
In this section we look at what is known about the distribution of residues within a single
period of F (mod m). That is, how frequently each residue is expected to appear. We start
off with a more specific question: for which moduli will all the residues appear with equal
frequency within a single period? If F (mod m) exhibits this property it is said to be uniform
or uniformly distributed. Kuipers and Shiue[12] proved that the only time F (mod m) can
possibly be uniform is when m = 5e .
It is not difficult to see that if F (mod m) is uniform, then necessarily m|k(m). We use
this requirement to prove that 5 is the only prime for which F (mod p) is uniform.
We first note that by inspection F (mod 2) is not uniform but F (mod 5) is, with each
residue appearing four times per period. Let us consider odd primes p.
If p = 5t 1 then F (mod p) cannot be uniform. For such p, we know k(p)|p 1, but
p6 | p 1 and so p6 | k(p).
If p = 5t 2 then F (mod p) cannot be uniform. In this case k(p)|2p + 2, but for odd p
we know p6 | 2p + 2. Hence p6 | k(p). Thus 5 is the only prime for which F (mod p) is uniform.
Now we mention a second fact about uniform distribution: if F (mod m) is uniform and
n|m, then F (mod n) is also uniform. Suppose m = nr. Then in one period of F (mod m)
each of the residues 0, 1,..., n 1 occurs with equal frequency as do the residues n, n + 1,...,
3.2. THE DISTRIBUTION OF RESIDUES
F(mod 5)
3
8
F(mod 25)
31
13
13
18 18 18
13
13
18
23 23 23
23
20 terms
100 terms
Figure 3.1: Comparing residues mod 5 and mod 25. The residues mod 25 are stratified
to display them more clearly.
2n 1, and so on until (r 1)n, (r 1)n + 1,..., rn 1. Thus within k(m) terms of F (mod n)
the residues 0, 1, 2, . . ., (n 1) appear with equal frequency, say t times. Since k(n)|k(m)
we can find these k(m) terms by simply repeating the first k(n) terms. If
within k(n) terms each residue appears
t
u
k(m)
k(n)
= u, then
times. Hence F (mod n) is uniform.
Putting these two facts together provides us with the following theorem.
Theorem 3.21 The only possible values of m for which F (mod m) is uniform are m = 5e .
It takes a very clever proof to finally settle the issue as we will demonstrate next. Niederreiter provided such a proof in [14].
Theorem 3.22 F (mod 5e ) is uniform for all integers e 1.
Before we start in earnest, we present the basic idea behind the proof. We know that
k(5e ) = 4 5e and so it will suffice to show that among the first 4 5e elements of the
sequence, we find exactly four elements, or, equivalently, at most four elements from each
residue class mod 5e .
For example, modulo 5 we see that the residue 3 appears exactly four times per period.
Modulo 25, the residues 3, 8, 13, 18, and 23 each appear exactly four times in one period.
Each of the five sections in figure 3.1 represents one period mod 5, and the bottom portion
shows where the residues 3, 8, 13, 18, and 23 appear in relation to the threes.
Proof: Let us assume that F (mod 5r1 ) is uniform and we will use induction
on r. We have seen that the hypothesis holds for r = 2.
32
Consider integers a and b such that 0 a < 5r1 and 0 b < 5r . (In the
figure we show a = 3 and r = 2.) Fn a (mod 5r1 ) has four solutions: n = c1 ,
c2 , c3 , c4 , with 0 c1 , c2 , c3 , c4 < 4 5r1 . Suppose b a (mod 5r1 ) and n = d
is a solution to Fn b (mod 5r ) where 0 d < 4 5r . Then Fd a (mod 5r1 ).
Hence by periodicity we must have d ci (mod 4 5r1 ) for some i. We will
show that there is only one solution d such that d ci (mod 4 5r ) for each i.
Suppose m n (mod 4 5r1 ) and Fm Fn (mod 5r ). Let 0 m, n < 4 5r ,
and without loss of generality, say m n. We will prove that, in fact, m = n.
From identity 2.2, Fn Fm (mod 5r ) implies
X n
X m
j
nm
5
2
5j
2j
+
1
2j + 1
j=0
j=0
(mod 5r ).
Now 45r1 |nm and by the Euler-Fermat theorem, (5r ) = 5r 5r1 = 45r1 .
Thus 2nm 1 (mod 5r ). Hence we have,
X n m
0
5j
2j + 1
2j + 1
j=0
Here we use the known identity
P2j+1 nm m
i=0
i
2j+1i . That is,

n
2j + 1

=
m
2j + 1
s+t
u

+
Pu
2j+1
X
i=1
i=0
s
i
nm
i

(mod 5r )
t
ui

(3.4)
to obtain
m
2j + 1 i
n
2j+1

.
Substituting this result into equation (3.4) gives us

"
X 2j+1
X
j=0
i=1
nm
i

m
2j + 1 i
#
0
(mod 5r ).
We claim that for j 1 each term in brackets is divisible by 5r . Consider

5j (nm)!
5j nm
= i!(nmi)!
. Cancelling out 5s from i! against 5j always leaves at
i
least one power of 5 in the latter. This is so since the largest value for i is 2j + 1
and the largest exponent t such that 5t divides (2j + 1)! is given by
t=
X
X
2j + 1
2j + 1
2j + 1
b
c
<
=
< j.
i
i
5
5
4
i=1
i=1
3.2. THE DISTRIBUTION OF RESIDUES
33
(The above identity can be found in [4]).

Since there is a factor of 5r1 in n m we get the desired divisibility
property.
Hence only the term corresponding to j = 0 remains.
That is,
n m 0 (mod 5r ). Now we can say that 5r |n m and 4 5r1 |n m. Thus

4 5r |n m. We know that 0 m, n < 4 5r , and so we must conclude that
n = m.
Therefore any residue mod 5r appears at most four times, which implies
that each residue must appear exactly four times, which gives us F (mod 5r ) is
uniform. The induction is complete and the theorem proved.
While F (mod m) is uniformly distributed only for m = 5e , one wonders if there is at least
a predictable distribution of residues under different moduli. In 1992 E. T. Jacobson[10]
fully described the distribution of residues in F (mod 2e ) and F (mod 2i 5j ) for i 5, and
j 0. We present his results below without proof.
Let v(m, b) be the number of occurrences of b as a residue in one period of F (mod m).
We have seen that v(5e , b) = 4 for all b (mod 5e ).
Theorem 3.23 For F (mod 2e ) the following is true:
For 1 e 4:
v(2, 0) = 1
v(2, 1) = 2
v(4, 0) = v(4, 2) = 1
v(8, 0) = v(8, 2) = v(16, 0) = v(16, 8) = 2
v(16, 2) = 4
v(2e , b) = 1 if b 3 (mod 4) and 2 e 4
v(2e , b) = 3 if b 1 (mod 4) and 2 e 4
v(2e , b) = 0 in all other cases.
For e 5:
1, if b 3 (mod 4)
2, if b 0 (mod 8)
3, if b 1 (mod 4)
v(2e , b) =
8, if b 2 (mod 32)
0, for all other residues.

For F (mod 2i 5j ) where i 5, and j 0, the following is true:
34
1, if b 3 (mod 4)
2, if b 0 (mod 8)
3, if b 1 (mod 4)
v(2i 5j , b) =
8, if b 2 (mod 32)
0, for all other residues.

It does not appear that any other work has been published on the distribution of residues
for other moduli.
3.3
The Zeros Of F(mod m)
The previous section indicates that the problem of describing all the residues for a given
modulus can be quite difficult, but when we restrict our attention to the behavior of the
zeros, many fascinating relationships become readily apparent.
Let (m) denote the subscript of the first positive term of the Fibonacci sequence which
is divisible by m. Vinson calls this the restricted period of F (mod m). Since the subscripts
of the terms for which Fn 0 (mod m) form a simple arithmetic progression, it is clear that
(m)|k(m). In fact one sees without too much difficulty that (m)|n if and only if m|Fn .
We use this idea in the following theorem.
Theorem 3.24 (m)|(mn).

Proof: By definition, we know that mn|F(mn) . Clearly then m|F(mn) , and
so (m)|(mn).
Let s(m) be the least residue of F(m)+1 modulo m. Because of the relationship (F(m) , F(m)+1 ) =
s(m)(0, 1) (mod m) we call s(m) the multiplier of F (mod m).
Finally, let (m) denote the order of s(m) modulo m. In other words, s(m)(m)
1 (mod m), and if n < (m) then s(m)n 6 1 (mod m).
Note the table in appendix B which compares the values of m, k(m), (m), and (m).
The next theorem ties together these three functions nicely. Robinson[15] states and proves
it, but also mentions that is a well known property. The proof presented here is the authors.
Theorem 3.25 k(m) = (m)(m).
3.3. THE ZEROS OF F(MOD M)
35
Proof: Suppose that a single period of F (mod m) is partitioned into smaller,

finite subsequences A0 , A1 , A2 , . . . as shown below:
0 1 s1 0 s1 s2 0 s2 s3 0 s3 0 1
| {z } | {z } | {z }
A0
A1
(3.5)
A2
Each subsequence Ai has (m) terms, it contains exactly one zero, and s1 =
s(m).
Every subsequence Ai for i 1 is a multiple of A0 . More precisely, the
following congruences hold modulo m.
A1
s1 A0
A2
s2 A0
..
.
An1
An
sn1 A0
sn A0
..
.
Now the last term in An1 is sn , and the last term in A0 is s1 . Thus
sn
(sn1 ) s1
(sn2 ) s1 s1
(sn3 ) s1 s1 s1
..
.
sn1
with congruences modulo m. Since (m) is the order of s1 , sequence (3.5) can
be rewritten as
(m)1
0 1 0 s1 0 s21 0 s31 0 s1
0 1
Thus (m) can be interpreted in a different way: it is the number of zeros in

a single period of F (mod m). Clearly it follows that k(m) = (m)(m).
There is an identity which follows from the proof.
36
n
Fr (mod m).
Identity 3.26 Fn(m)+r F(m)+1
Proof: The identity comes from the fact that An sn1 A0 . More specifically,
the rth term of An is congruent to sn1 times the rth term of A0 modulo m.
As a point of interest, note the property that gcd(m, si ) = 1 for all i. This must be the
(m) i
case since (si )(m) = (si1 )(m) = (s1
) 1 (mod m). We could also draw this conclusion
from theorem 1.1, realizing that if m|Fn then m and Fn+1 have no nontrivial divisors in
common.
The following theorem doesnt appear to give us any immediate insight into F (mod m),
but a couple of nice corollaries follow from it. The proof comes from Robinson [15], but he
acknowledges that Morgan Wood knew the result in the early 1930s.
Theorem 3.27 k(m) = gcd(2, (m)) lcm[(m), (m)] where (2) = 1 and (m) = 2 for
m > 2.
Proof: By identity 1.8, Fn2 Fn+1 Fn1 = (1)n+1 so
2
F(m)
F(m)+1 F(m)1 = (1)(m)+1 .
Since F(m) 0 and F(m)+1 F(m)1 (mod m),

2
F(m)+1
(1)(m)+1
(mod m).
That is,
(s(m))2 (1)(m)
(mod m).
Thus (s(m))2 and (1)(m) have the same order modulo m. Specifically,
(m)
(m)
=
gcd(2, (m))
gcd((m), (m))
where (m) is the order of 1 modulo m. Thus
k(m)
= (m)(m) = (m)
gcd(2, (m)) (m)

gcd((m), (m))
gcd(2, (m)) lcm[(m), (m)].
Corollary 3.28 k(m) is even for m > 2.
(3.6)
37
Proof: By the proceeding theorem, if k(m) is odd we must have lcm[(m), (m)]
odd. For that to happen (m) must be odd. By the nature of (m) then m = 2.
Hence the contrapositive (for applicable values of m): If m > 2 then k(m) is
even.
One of the more surprising properties of F (mod m) is demonstrated in the following
corollary.
Corollary 3.29 (m) = 1, 2, or 4.
Proof:
k(m)
gcd(2, (m)) lcm[(m), (m)]
(1 or 2) ((m) or 2(m))
= (m), 2(m), or 4(m).

Therefore (m) = 1, 2, or 4.
Recall that it is this theorem which allows us to say that k(m) < m2 1 for m 6.
We have already seen that (m) shares a property with k(m) in that (m)|(mn). We
will demonstrate that some other properties of k(m) are also exhibited in (m). Like k(m),
Q ei
Vinson[21] shows we can express (m) in terms of (pei i ) where m =
pi is the prime
factorization of m.
Theorem 3.30 If m has the prime factorization m =
pei i then (m) = lcm[(pei i )].
Proof: Notice,
m|Fn
pei i |Fn for all i
(pei i )|n for all i.
The smallest n which satisfies the last condition, and hence all of them, is n =
lcm[(pei i )]. Thus according to the first condition, Fn is the smallest Fibonacci
number divisible by m. That is, (m) = n = lcm[(pei i )].
Again, like k(m), we can express (m) in a slightly more convenient form.
38
Theorem 3.31 ([m, n]) = [(m), (n)], where brackets denote the least common multiple
function.
Proof:
Ft 0 (mod [m, n])
Ft 0 (mod m) and Ft 0 (mod n)

(m)|t and (n)|t
The smallest t for which the first condition is true is t = ([m, n]). the smallest
t for which the last condition is true is t = [(m), (n)]. The theorem follows.
A third similarity exists: if p is an odd prime and t is the largest integer such that
(pt ) = (p) then (pe ) = pet (p). This result exists as a corollary to yet another
surprising theorem from [15].
Theorem 3.32 For p any odd prime, (pe ) = (p).
Proof: We again make use of the Fibonacci matrix, U . By equation (3.3) we
know that
e+1
U (p
s(pe+1 )I (mod pe+1 ).
Furthermore,
e
U (p
s(pe )I (mod pe )
which implies
e
U p(p
(s(pe )I + pe B)p (s(pe ))p I (mod pe+1 ).
Hence (pe+1 )|p(pe ). Since (m)|(mn) we also know (pe )|(pe+1 ). Consequently, (pe+1 ) = (pe ) or p(pe ). It follows that
can say
k(p )
k(p)
(pe )
(p)
= pj . Clearly,
k(pe ) (pe )
k(pe ) k(p)
(p) (pe )
(p) k(p)
and so
(pe ) k(pe )
k(p) k(pe )
.
(p) (pe )
(p) k(p)
= pi and similarly we
Since
k(pe )
(pe )
= (pe ) and
k(p)
(p)
39
= (p),
pi (1, 2, or 4) = (1, 2, or 4)pj .

Since we assumed p to be odd, we must have pi = pj . That is,
which implies
k(p)
(p)
k(pe )
(pe ) .
(pe )
(p)
k(pe )
k(p)
In other words, (p) = (pe ).
Corollary 3.33 If p is an odd prime and t is the largest integer such that (pt ) = (p)
then (pe ) = pet (p). In fact, this t is also the largest integer such that k(pt ) = k(p).
Proof: This follows directly from the previous proof where
(pe )
(p)
k(pe )
k(p) .
The case where p = 2 exists as a corollary to theorem 3.36.

We will now examine the function (m) more closely. The next three theorems present
some useful relationships between (m), (m), and k(m).
Theorem 3.34 For m 3, (m) = 4 if and only if (m) is odd.
Proof: Assume (m) = 4. Then (s(m))2 is the residue after the second zero
and by equation (3.6) we know (s(m))2 (1)(m) . Clearly, (s(m))2 6 1 and
so (m) must be odd.
Assume (m) is odd. In this case, equation (3.6) tells us that (s(m))2 1.
The only possible value for (m) now is 4.
In order to show some necessary and sufficient conditions for when (m) = 1, we rely
on the identity Fk(m)j Fj = (1)j+1 Fj (mod m).
Theorem 3.35 (m) = 1 if and only if 46 | k(m).
Proof: We will prove the contrapositive of the theorem in both directions. First,
assume that 4|k(m). Then if we let j =
k(m)
2
+ 1 in the identity preceding the
theorem, and make note that this j is odd, we get F k(m) 1 F k(m) +1 (mod m).
2
By the Fibonacci recurrence relation this implies F k(m) 0 (mod m), which in
2
turn implies (m) 6= 1.

If (m) 6= 1 then (m) = 2 or 4. We know that k(m) = (m)(m), so if
(m) = 4 then clearly 4|k(m). If (m) = 2, then by the previous theorem (m)
is even, and once again 4|k(m).
40
Since theorems 3.34 and 3.35 are if and only if, the next theorem appears as a natural
consequence.
Theorem 3.36 (m) = 2 if and only if 4|k(m) and (m) is even.
Corollary 3.37 If 4|(m) then (m) = 2.
Corollary 3.38 If 8|k(m) then (m) = 2.
Corollary 3.39 (2) = (4) = 1. For e 3, (2e ) = 2.
Proof: By inspection, (2) = (4) = 1 and (8) = 2. From theorem 3.5 we
know that k(2e ) = 3 2e1 . Since k(16) = 24 we see 8|k(2e ) for e 4, and we
can apply the preceding corollary.
We must be careful when we try to apply theorem 3.36. We can not conclude that if
(m) = 2 then 4|(m). For example, (40) = 2, yet (40) = 30. However, when the
modulus is a prime or a power of a prime, theorem 3.36 can be strengthened. The following
theorem found in [21] will be used later.
Theorem 3.40 Let p be an odd prime. If (pe ) = 2 then 4|(pe ).
Proof:
First we show that the theorem is true for e = 1. When (p) = 2,
theorem 3.34 assures us that (p) is even. In identity 3.26, let n = 1 and
r = 12 (p) to attain
F 12 (p) F(p)+1 F 12 (p)
(mod p).
Noting that F(p)+1 = s(p) and that s(p)2 (1)(p) (mod p), we can multiply
both sides of the above congruence by F(p)+1 and apply identity 1.3 to achieve
1
F(p)+1 F 12 (p) (1)(p) (1) 2 (p)+1 F 12 (p)
(mod p).
We recall that (p) is even, then multiply both sides by F 1p2

and apply Fer(p)
2
mats theorem to get

1
F(p)+1 (1) 2 (p)+1
(mod p).
41
Since (p) = 2 we know that F(p)+1 6 1. Thus (1) 2 (p)+1 = 1 which implies
4|(p) and we have proved the theorem for e = 1.
Since p is odd, (pe ) = 2 implies that (p) = 2. We have seen that this
implies 4|(p), and so by theorem 3.24, 4|(pe ). Thus the theorem is proved.
So far we have been able to find (m) only by analyzing k(m) or (m). The next theorem
gives us a method for finding ([m, n]) if we know (m) and (n). Vinson does not state
this theorem explicitly, but the author was able to construct it from the information given
by him.
Theorem 3.41 Given (m) and (n), the table below determines the value of ([m, n]).
Once again, brackets denote the least common multiple function.
([m, n])
(n)
(m)
4 if m = 2
2 otherwise
4 if n = 2
2 otherwise
42
Proof: Let
(m) = 2r a
(n) = 2w b
(m) = 2s
(n) = 2x
k(m)2t a
k(n)2y b
where a, b are odd integers. Hence

[k(m), k(n)] = 2max(t,y) [a, b]
and
[(m), (n)] = 2max(r,w) [a, b].
Then
([m, n]) =
[k(m), k(n)]
k([m, n])
=
= 2max(t,y)max(r,w) .
([m, n])
[(m), (n)]
Notice that r + s = t and w + x = y. We now address four cases.

Case 1: To describe the diagonal of the table, suppose (m) = (n), that is,
s = x. Then max(t, y) max(r, w) = max(r + s, w + x) max(r, w) = s = x.
Hence ([m, n]) = 2s = 2x = (m) = (n).
In order to prove the remaining three cases it is helpful to translate theorems 3.35, 3.36, and 3.34 into statements about r, s, and t:
If s = 0 then t = 0 (for m = 2) or t = 1 (for m > 2).
Respectively, r = 0 or r = 1.
If s = 1 then t 2 and r 1.
If s = 2 then r = 0 and hence, t = 2.
Analogous statements hold for w, x, and y.
Case 2: Suppose (m) = 4 and (n) = 1, that is, s = 2 and x = 0. Since
s = 2 we know that r = 0 and t = 2. Also, since x = 0 we know that either
w = 0 or w = 1.
Case 2a. Say x = 0, w = 0, and so y = 0. Then max(t, y) max(r, w) =
2 0 = 2. Hence ([m, n]) = 22 = 4.
Case 2b. Say x = 0, w = 1, and so y = 1. Then max(t, y) max(r, w) =
2 1 = 1. Hence ([m, n]) = 21 = 2.
Case 3: Suppose (m) = 2 and (n) = 1, that is, s = 1 and x = 0. Since
s = 1 we know that t 2 and r 1. Again, since x = 0 either w = 0 or w = 1.
Case 3a. Say x = 0, w = 0, and so y = 0. Then max(t, y) max(r, w) =
t r = s = 1. Hence ([m, n]) = 21 = 2.
43
Case 3b. Say x = 0, w = 1, and so y = 1. Then max(t, y) max(r, w) =

t r = s = 1. Hence ([m, n]) = 21 = 2.
Case 4: Suppose (m) = 4 and (n) = 2, that is, s = 2 and x = 1. Then
r = 0, t = 2, and w 1, y 2. Then max(t, y) max(r, w) = y w = x = 1.
Hence ([m, n]) = 21 = 2.
This completes the proof.
There are two interesting corollaries that follow.
Corollary 3.42 If 3|m then (m) = 2.
Proof: Since (3) = 2 we know that (3e ) = 2. Let m be expressed as 3e n
where 36 | n. By the previous theorem, then (m) = ([3e , n]) = 2.
Corollary 3.43 (m) = 1 if and only if 86 | m and (p) 2 (mod 4) for all odd primes p
that divide m.
Proof: Let m have the prime factorization m =
pei i . Suppose (m) = 1. By
theorem 3.41 it is easy to see that we must have (pei i ) = 1 for all i. If p1 is the
smallest prime in the prime factorization of m and p1 = 2 then by corollary 3.39,
e1 2 and thus 86 | m. Recall that for odd p, (pei i ) = (pi ). Theorem 3.36
tells us that if (pi ) 1 or 3 (mod 4) then (pi ) = 4. Theorem 3.34 tells us
that if (pi ) 0 (mod 4) then (pi ) = 4. Hence when (m) = 1 we must have
(pi ) 2 (mod 4) for all i.
Suppose that 86 | m. Thus if 2e is a factor of m, then e 2 and (2e ) = 1.
For odd p, if (p) 2 (mod 4) then (p) = 1 by theorems 3.35 and 3.40. Hence
if we suppose that (pei i ) 2 (mod 4) for all i we have (pei i ) = 1 for all i and
then theorem 3.41 indicates that (m) = 1.
We now know that for composite m, (m) can be determined by factoring m into smaller
moduli. We also know that for odd primes p, (pe ) = (p). Can we determine (p) for
odd primes p? While k(p) and (p) have strongly resisted this analysis, we can make some
progress on (p). The four results are grouped together in the following theorem found in
[21].
44
Theorem 3.44 For odd primes p,

(i)
(ii)
If p 11 or 19 (mod 20) then (pe ) = 1.

If p 3 or 7 (mod 20) then (pe ) = 2.
(iii)
If p 13 or 17 (mod 20) then (pe ) = 4.
(iv)
If p 21 or 29 (mod 40) then (pe ) 6= 2.
Proof: Since we are assuming p odd, we need only prove each part for e = 1
and the conclusion will follow. Before looking at the individual cases we take
some time to develop a couple useful facts. First, since (p) is the order of s(p),
we know that (s(p))(p) 1 (mod p). In addition, we know by Fermats theorem
that (s(p))p1 1 (mod p). Thus (p)|p 1 and it follows that if p 3 (mod 4)
then (p) 6= 4.
Secondly, if p = 5t 2 then Fp 1 and Fp+1 0 (mod p). We see that
(p)|p + 1 but k(p)6 | p + 1. Hence (p) 6= k(p) and consequently (p) 6= 1. Now
we turn our attention to the four results.
(i) Here p 3 (mod 4), so (p) 6= 4. Assume (p) = 2. Then by theorem 3.36, 4|k(p). Since p = 5t 1 we have k(p)|p 1, and so 4|p 1. However,
this is a contradiction since p 1 10 or 18 (mod 20). Thus (p) = 1.
(ii) Again, p 3 (mod 4), so (p) 6= 4. Now p = 5t 2 and we have seen
that this implies (p) 6= 1. Thus (p) = 2.
(iii) As in the previous part, p = 5t 2 and so (p) 6= 1. Also, we know
(p)|p + 1. In this case p 1 (mod 4) so 46 | p + 1. Hence 46 | (p). Applying
theorem 3.40 we have (p) 6= 2. Hence (p) = 4.
(iv) Suppose (p) = 2. By theorem 3.40, 4|(p) and so consequently 8|k(p).
Furthermore, since p = 5t 1, theorem 3.12 assures us that k(p)|p 1. It follows
that 8|p 1. However, p 1 20 or 28 (mod 40) so clearly 86 | p 1. Thus we
have our contradiction and we can say (p) 6= 2.
We would like to know if anything can be said about the primes not covered by the
theorem, namely p 1 or 9 (mod 40). Also, can we be more exact about primes where
p 21 or 29 (mod 40)? Vinson provides the following examples to show that his theorem
45
is complete:
For p
1 (mod 40) :
(521) = 1,
(41) = 2,
(761) = 4
For p
9 (mod 40) :
(809) = 1,
(409) = 2,
(89) = 4
For p
21 (mod 40) :
(101) = 1,
(61) = 4
For p
29 (mod 40) :
(29) = 1,
(109) = 4
We finish section 3.3 with a couple of theorems promised at the end of section 3.1. While
they speak to the character of the period, their proofs are made easy by the theory developed
in this section.
Theorem 3.45
(i) If n 5 is odd then k(Fn ) = 4n.

(ii) If n 4 is even then k(Fn ) = 2n.
Proof: We first note that if Fn is used for the modulus, then naturally (Fn ) =
n.
(i) This result follows from the fact that for m 3, (m) odd implies (m) =
4. For n 5 and odd, Fn 3 and (Fn ) = n is odd. Thus k(Fn ) = 4n.
(ii) Here, (Fn ) = n is even so (Fn ) = 1 or 2. If (Fn ) = 1 then Fn1
Fn+1 1 (mod Fn ). However, F1 , F2 , F3 , . . . , Fn1 is non decreasing and F3
2 (mod Fn ), hence our contradiction. Therefore, (Fn ) = 2 and k(Fn ) = 2n.
Theorem 3.46
(i) If n 3 is odd then k(Ln ) = 2n.

(ii) If n 2 is even then k(Ln ) = 4n.
Proof: First we establish that (Ln ) = 2n. From identity 1.10, Fn Ln = F2n
which implies F2n 0 (mod Ln ) so certainly (Ln )|2n. Now if (Ln ) 6= 2n
then (Ln ) n. In other words, Ln |Ft for some t n. However, Ln > Ft for
2 t n, so clearly Ln6 | Ft . Hence (Ln ) = 2n.
(i)
Fn+1 Ln
= Fn+1 (Fn+1 + Fn1 )

2
= Fn+1
+ Fn+1 Fn1
2
= Fn+1
+ Fn2 + (1)n (by identity 1.8)
= F2n+1 1 (by identity 1.7 and n odd)
46
Thus F2n+1 1 (mod Ln ) and we can say (Ln ) = 1. Therefore k(Ln ) =

(Ln ) = 2n.
(ii) When n is even, 4|(Ln ) and so by corollary 3.37, (Ln ) = 2. Hence
k(Ln ) = (Ln ) 2 = 4n.
Chapter 4
Personal Findings
During my survey of known Fibonacci properties I was fortunate enough to stumble across
a few areas where apparently very little research had been done. Upon examining these
areas more closely I was able to discover yet more astounding properties of the Fibonacci
sequence.
4.1
Spirolaterals And The Fibonacci Sequence
Spirolaterals, invented in 1973 by Frank C. Odds, are simple graphical representations of

finite integer sequences. The rules for creating a spirolateral from a given sequence are
simple, and the spirolateral can easily be drawn on a sheet of ordinary graph paper. Suppose
we have a sequence x1 , x2 , x3 , . . . , xn . To create the spirolateral draw a line from left to
right x1 units long, turn right 90 , draw a line x2 units long, turn right 90 and continue
in this manner. When the end of the sequence has been reached, start over again with x1 .
Figure 4.1: Some simple spirolaterals. The circle indicates the starting point.
Eventually, either the line will return to its starting point heading in the initial direction, or
else the pattern will wander off the page. It is known that when n 1 or 3 (mod 4) then the
spirolateral exhibits 4-fold symmetry. When n 2 (mod 4) then then spirolateral exhibits
2-fold symmetry, and when n 0 (mod 4) the spirolateral does not exhibit symmetry and
usually glides off the page in some diagonal direction.
I wrote a short computer program that randomly generated sequences and then drew
47
48
CHAPTER 4. PERSONAL FINDINGS
them as spirolaterals. After playing with the spirolateral program for a while, viewing
hundreds of spirolaterals, I was accustomed to seeing interesting symmetries and curious
patterns fly off the screen at some diagonal. Then, when I used a period of residues of
the Fibonacci sequence under various moduli to generate spirolaterals I was surprised that
typically the results were not symmetric and often did not translate off the screen.
Figure 4.2: Some spirolaterals using Fibonacci residues. Here, F (mod 5) and F (mod 8) are
both asymmetric and nontranslating. Circles indicate starting points.
Apparently, after only one time through the period, the line had returned to the starting
point and was heading in the initial direction. That is, the sum of the lines drawn to the
right equaled the sum of the lines drawn to the left, and the sum of the lines drawn up
equaled the sum of the lines drawn down. This seemed a fairly remarkable occurrence, for
it indicated a truth about the alternating sum of the residues themselves. After studying
many examples, a conjecture was made and eventually a proof was given showing when the
alternating sum will be zero.
In order to express clearly the idea of working with the residues themselves, let us
introduce some notation. Let fn represent the least nonnegative residue of Fn modulo m.
As we would expect, fn = fn+rk(m) for any integer r. When n is odd, Fn = Fn and so
fn = fn . When n is even, Fn = Fn which implies Fn + Fn 0 (mod m) and so either
fn = fn = 0 or else fn + fn = m
Theorem 4.1 4|k(m) if and only if
k(m)
2 1
(1)i f2i+1 = 0.
i=0
Proof: Let k = k(m) and rewrite the above summation as

f1 f3 + f5 fk5 + fk3 fk1
=
(f1 fk1 ) (f3 fk3 ) + (f k 1 f k +1 ).

2
4.1. SPIROLATERALS AND THE FIBONACCI SEQUENCE
49
First suppose that 4|k. When n is odd fn = fn = fkn . Hence each term in
parentheses equals zero, implying the entire summation equals zero.
On the other hand, since fn = fkn for odd n we know that each parenthesized term must equal zero, and in particular f k 1 = f k +1 . By the recurrence
2
relation we know then f k = 0. Clearly now (m) 6= 1 and by theorem 3.35 we

2
can say 4|k.

It turns out that the same conditions do not ensure that the alternating sum of the
evenly subscripted residues will be zero. However, the conditions needed in this case are
not very different. First, though, we need a lemma.
Lemma 4.2 Suppose 4|k(m) for some m and let j be even. If j is a multiple of
k(m)
2
fj = fkj = 0 otherwise fj + fkj = m.

Proof:
Let k = k(m) and take all congruences (mod m). When j is even
Fkj (1)j+1 Fj Fj . Thus if Fj 6 0 then fj + fkj = m. Since 4|k we

know by theorem 3.35 that F k 0 and consequently if j is a multiple of
2
fj = 0. Also, k j will be a multiple of
k
2
k
2
then
so fkj = 0. Suppose Ft 0 for some
0 < t < k2 . Then (m) = 4 and (m) = t is necessarily odd. Thus the only time
Fj 0 and j is even is when j is a multiple of
k
2.
The lemma follows.
Theorem 4.3 If k(m) 4 (mod 8) then

k(m)
2 1
(1)i f2i = 0.
i=0
Proof: The summation above is

f0 f2 + f4 f6 + + fk4 fk2
= f0 (f2 + fk2 ) + (f4 + fk4 ) (f k 2 + f k +2 ) + f k .
2
By our lemma f0 = f k = 0 and each quantity in parentheses equals m. There

2
are
k
4
1 k
2(2
2) =
k
4
1 parenthesized terms and since k 4 (mod 8) we know
1 is even. Hence the entire summation is zero.
then
50
We can extend our idea of the spirolateral and instead of restricting ourselves to just 90
turns we can make spirolaterals with 60 turns or 45 etc... If 60 turns are used, hexagonal
type patterns emerge. Occasionally these pattern will return to their starting place when
F (mod m) for some m is used as the generating sequence. Since every third line is parallel
in these pictures, this result indicates that the alternating sum of every third term in these
sequences is zero, regardless of where we start to take our sum. The following conjecture
expresses this notion.
Conjecture 4.4 If k(m) 12 (mod 24) and (m) = 4 for some m, then
k(m)/31
(1)i f3i+j = 0
i=0
for j = 0, 1, 2.
After only a cursory investigation it appears that this conjecture may yield to a proof if
given a couple hours of thought. Also, many times the alternating sum of every fourth, fifth,
and so on, residue in a period equals zero. This area seems to be quite open for research.
After trying for a while to find results pertaining to the summation of the residues
themselves I turned to a related question. If I take the sum (or alternating sum) of every
nth term of the Fibonacci sequence within a period of F (mod m), what will the result be,
modulo m? For example, k(5) = 20, so what is the sum modulo 5 of every fourth term
starting with, say, F2 ? Or what can I expect of F3 F8 + F13 F18 (mod 5)? To this end
the following two identities are vital. The following identity is due to Siler[16].
Identity 4.5 For 0 j < n and t 0, we have
t
X
i=0
Fni+j =
Fnt+n+j Fj + (1)j+1 Fnj + (1)n+1 Fnt+j

.
Ln 1 + (1)n+1
Proof: The identity = 1 will be used several times in the proof.

t
X
i=0
Fni+j
t
X
1
( ni+j ni+j )
5
i=0
51
"
#
t
t
X
X
1
j
n i
j
n i

( )
( )
5
i=0
i=0

1 j (1 ( n )t+1 ) j (1 ( n )t+1 )
1 n
1 n
5

1 j (1 n ) nt+n+j (1 n ) j (1 n ) + nt+n+j (1 n )
(1 n )(1 n )
5

1 ( j j ) + (1)j ( nj nj ) ( nt+n+j nt+n+j ) + (1)n ( nt+j nt+j )
1 + ( )n ( n + n )
5
Fj + (1)j Fnj Fnt+n+j + (1)n Fnt+j

.
1 + (1)n Ln
By multiplying the numerator and denominator by 1 we obtain the identity.
The second identity is similar, but concerns alternating sums. I have been unable to find
this identity published anywhere, and the proof below is due to Fredric Howard.
Identity 4.6 For 0 j < n and t 0, we have
t
X
(1)i Fni+j =
i=0
Fj + (1)j+1 Fnj + (1)t Fnt+n+j + (1)t+n Fnt+j

.
1 + (1)n + Ln
Proof:
t
X
i=0
(1)i Fni+j
t
X
(1)i ni+j
(
ni+j )
5
i=0
"
#
t
t
X
X
1
j
n i
j
n i

( )
( )
5
i=0
i=0

1 j (1 ( n )t+1 ) j (1 ( n )t+1 )
1 ( n )
1 ( n )
5

1 j (1 + n ) + (1)t nt+n+j (1 + n ) j (1 + n ) (1)t nt+n+j (1 + n )
(1 + n )(1 + n )
5
52
( j j ) + (1)j+1 ( nj nj ) +
(1)t ( nt+n+j nt+n+j ) + (1)t+n ( nt+j nt+j )
1
1 + ( )n + ( n + n )
5
Fj + (1)j+1 Fnj + (1)t Fnt+n+j + (1)t+n Fnt+j

.
1 + (1)n + Ln
If we fix n and j and sum up (using either identity) all terms of the form Fni+j within
a single period of the Fibonacci sequence (F0 Fni+j < Fk(m) ) what will the result be? In
particular, for what values of n, j, and m will the sum be congruent to zero modulo m?
We will only consider those n such that n|k(m). We are not especially interested in
looking at, say, every seventh term if the period is not a multiple of seven. When t =
+/
let Sn denote the summation of identity 4.5 and let Sn
k(m)
n 1
denote the alternating summation
of identity 4.6.
Hence, letting k = k(m),
k
n 1
+
Sn =
Fni+j =
i=0
Fk+j Fj + (1)j+1 Fnj + (1)n+1 Fkn+j

.
Ln 1 + (1)n+1
Since Fj Fk+j (mod m),

+
Sn (Ln 1 + (1)n+1 ) (1)j+1 Fnj + (1)n+1 Fk(nj)

(1)j+1 Fnj + (1)n+1+(nj)+1 Fnj
0
(mod m).
Quite surprisingly, j has dropped out of our equation and the following is then true for all
j:
+
Theorem 4.7 If gcd(m, Ln 1 + (1)n+1 ) = 1 then Sn 0 (mod m).

We will now look at some examples to demonstrate the surprising results this theorem
gives us.
+
When n = 1, Ln 1 + (1)n+1 = 1, and so for any modulus m, S1 0 (mod m). In

other words, F0 + F1 + F2 + + Fk(m)1 0 (mod m) for all m.
+
When n = 2, Ln 1 + (1)n+1 = 1, and so for any modulus m, S2 0 (mod m).

Here Ln 1 + (1)n+1 = 5. Thus if 4|k(m) but 5 6 | m then
Pk/41
Pk/41
Pk/41
0 (mod m). Recall that this means i=0 F4i i=0 F4i+1 i=0 F4i+2
Consider n = 4.
+
S4
Pk/41
i=0
53
F4i+3 0 (mod m). Again, it does seem remarkable that the point where we start
to take every fourth term does not matter at all. Finding some values for m which satisfy the
above conditions is easy to do. By inspection we look for an m such that 4|k(m) then using
the fact that k(m)|k(mr) we let r be any integer except multiples of 5. We see k(3) = 8 so
some viable choices for m are 3, 6, 9, 12, 18, 21, 24, 27, 33, ... Notice the omission of 15
and 30. Of course, these arent the only possible values for m. We find k(7) = 16, so 7, 14,
21, 28, 42, ... are all valid choices as well. Amazingly, for all these values of m (and many
+
more), S4 0 (mod m).

+
Next we fix m = 17 and find all n such that Sn 0 (mod 17). Now k(17) = 36 so we will
only consider those n which divide 36. It just happens that gcd(17, Ln 1 + (1)n+1 ) = 1
+
for all those n and so Sn 0 (mod 17) for n = 1, 2, 3, 4, 6, 9, 12, and 18. Truly amazing!
When m = 9, k(m) = 24. We find that gcd(9, Ln 1 + (1)n+1 ) = 1 for all divisors, n,
+
of 24 except n = 8. Hence Sn 0 (mod 9) for n = 1, 2, 3, 4, 6, and 12.

It should be noted that we need not always use the first k(m) terms of the Fibonacci
sequence in our summations. This comes from the fact that we are summing over a single
period of the Fibonacci sequence and the value of j in identities 4.5 and 4.6 is inconsequential.
Considering the last example, we can take any 24 consecutive Fibonacci numbers, then add
up, say, every third, and our total will continue to be a multiple of 9.
+/
Now we turn our attention to the alternating sum, Sn . Again we want

an integer, but two cases arise:
k(m)
n
is either even or odd. The case where
k(m)
n
k(m)
n
to be
is even is
explored here. In this case the number of positive terms in the summation is the same as the
number of negative terms. The case where
k(m)
n
is odd does not appear to give nice results
(probably because it lacks the symmetry of the first case) and has not been explored a
great deal.
+/
As before, we will now try to find out when Sn
0 (mod m). Let k = k(m). By
identity 4.6:
+/
Sn
k
n 1
X
i=0
Assuming
k
n
(1)i Fni+j =
Fj + (1)j+1 Fnj + (1) n 1 Fk+j + (1) n 1+n Fkn+j

.
Ln + 1 + (1)n
is even, and noting Fj Fk+j (mod m),
+/
Sn (Ln + 1 + (1)n ) (1)j+1 Fnj + (1) n 1+n (1)nj+1 Fnj
54
(1)j+1 Fnj + (1)j Fnj

0
(mod m)
Once again j has dropped out and we are left with the following result.
Theorem 4.8 If
k(m)
n
+/
is even and gcd(m, Ln + 1 + (1)n ) = 1 then Sn
0 (mod m).
Here are a couple examples of the application of this theorem.

k(m)
4
Let n = 4. Then Ln + 1 + (1)n = 9. Now we want the values for m such that
is even and gcd(m, 9) = 1, the latter requirement being the same as 3 6 | m. After some
inspection we find m = 7 works since k(m) = 16. Thus m = 7, 14, 28, 35, 49, ... are all
acceptable values. Also, k(16) = 24 so m = 16, 32, 64, 80, 112, ... work as well. For all
+/
these values of m, S4
0 (mod m).
When m = 9, k(m) = 24. We see that
k(m)
n
is even for n = 1, 2, 3, 4, 6, 12. Of these

+/
values, gcd(9, Ln + 1 + (1)n ) = 1 for n = 1, 2, 3, 6. Hence S1
+/
S2
+/
S3
+/
S6
0 (mod 9).
When m = 17, k(m) = 36. Now
k(m)
n
is even for n = 1, 2, 3, 6, 9, 18. Of these values,

+/
gcd(17, Ln + 1 + (1)n ) = 1 for n = 1, 2, 3, 6, 9. Hence S1

+/
S9
+/
S2
+/
S3
+/
S6
0 (mod 17).
It is remarkable that in the example above where m = 9, the alternating sum of the
residues themselves for n = 1, 2, 3, and 6 actually equals zero when j 6= 0. When j = 0, the
alternating sum of the residues equals 9 for all four values of n.
Similarly, in the example when m = 17, the alternating sum of the residues themselves
actually equals zero for all j. Clearly this is an area that is wide open for research and
appears to have some fascinating results.
Though we have provided some sufficient conditions indicating when a sum will be
congruent to zero, these conditions are nowhere close to necessary. There are many times
when a sum will be congruent to zero for some j but not all. It appears that with some
work, new and more general sufficient conditions may be found.
The notion of summing Fibonacci numbers over a single period is not new and has been
examined previously by Aydin and Smith in [3]. Their approach, however, was to take the
sum of powers of all the Fibonacci numbers in a period of F (mod p) and observe its value
4.2. FIBONACCI SUBSEQUENCES
For all primes p
55
P
P Fi2 0
P Fi3 0
Fi 0
P
(1)i Fi 0
P
(1)i Fi3 0
P
(1)i Fi4 0
For p 6= 3
For p 6= 11
Fi5 0
P
(1)i Fi5 0
For p 6= 11, 29
P 6
P Fi7 0
Fi 0
P
(1)i Fi7 0
Table 4.1: Some sums of powers of Fibonacci numbers.

modulo p, where p is a prime. Table 4.1 displays a few of their results. All sums are taken
over a single period and all congruences are (mod p).
It seems that the next step is to look at summations of the form
e
Fni+j
and
P
e
(1)i Fni+j
.
While no known work has been done in this area, empirical evidence suggests that there are
many interesting theorems waiting to be proven.
4.2
Fibonacci Subsequences
During my research I came across a paper [8] by Herta Freitag in which she describes
a property of the unit digits of the Fibonacci numbers. In the paper she examines several subsequences of F (mod 10) where the terms of the subsequence actually follow the
usual Fibonacci recurrence relation. First she conjectures and proves that Fn + Fn+5
Fn+10 (mod 10) for all n, then she shows that if j {1, 5, 13, 17, 25, 29, 37, 41, 49, 53} the
relation Fn + Fn+j Fn+2j (mod 10) still holds for all n.
Her paper leaves many questions open. Are there other subsequences of this type in
F (mod 10) which do depend on the value of n as defined above? What can we say about the
subsequences of F (mod m) for arbitrary m? We can immediately answer the first question
by noting that F0 , F9 , F18 , . . . (mod 10) forms such a sequence, but F1 , F10 , F19 , . . . (mod 10)
does not. We will examine the second question closely.
First we need some notation. If a subsequence H of F (mod m) exhibits the recurrence
56
relation Hn+2 Hn+1 + Hn (mod m), let us call H a Fibonacci subsequence modulo
m. We will drop the modulo m when the modulus is understood. We will only consider
subsequences whose terms are evenly spaced throughout F (mod m) and we will denote such
subsequence by {Fn , Fn+j } where Fn and Fn+j are successive term of the subsequence. We
will often use the variables n and j like this where Fn is a term in the subsequence and j is
the spread of the subsequence. Since F (mod m) has period k(m), we will always assume
0 n < k(m) and 1 j k(m).
Let us examine some properties of these Fibonacci subsequences. Let d =
gcd(j, k(m)). If {Fn , Fn+j } is a Fibonacci subsequence then {Fn+dx , Fn+dx+j } is also a
Fibonacci subsequence for all x. To see this, consider a period of F (mod m) where we start
at Fn then take every j th term. When we reach the end of the period, we loop back to the
start and continue. After wrapping around once or several times we will eventually return
to Fn . In the process, we will have taken every dth number in the period. Thus not only
is {Fn+dx , Fn+dx+j } a Fibonacci subsequence, in a sense it is the same subsequence with a
different starting point. We say that it is a rotation of the subsequence {Fn , Fn+j }.
In particular, notice that if {Fn , Fn+j } is a Fibonacci subsequence and gcd(j, k(m)) = 1,
then {Ft , Ft+j } is a rotation of {Fn , Fn+j } for any t, and the subsequence has k(m) terms.
So how do we find these Fibonacci subsequences? Given only a few successive terms of
{Fn , Fn+j }, can we say whether or not it is a Fibonacci subsequence, without having to
compute the entire subsequence? As a matter of fact, the main result of this section proves
that in many cases we can do just that. Before we present the main result we provide the
following two lemmas.
Lemma 4.9 {F0 , Fj } is a Fibonacci subsequence if and only if {Fn , Fn+j } is a Fibonacci
subsequence with Fn 0.
Proof:
rem 3.25.
The ideas in this proof are similar to those in the proof of theoLet s be the residue of Fn+1 (mod m).
The pair (Fn , Fn+1 )
s(F0 , F1 ) (mod m), hence the sequence Fn , Fn+1 , Fn+2 , . . . (mod m) is the same
as the sequence sF0 , sF1 , sF2 , . . . (mod m). Certainly, if {F0 , Fj } is a Fibonacci
subsequence then {sF0 , sFj } is a Fibonacci subsequence, and hence {Fn , Fn+j }
is a Fibonacci subsequence.
57
In the other direction, we know that gcd(s, m) = 1, (this is seen in the

remark after identity 3.26) therefore there exists a t such that t(Fn , Fn+1 )
(F0 , F1 ) (mod m), and the result follows as before.
Lemma 4.10 For odd j, Fnj +Fn Fn+j (mod m) if and only if (Lj 1)Fn 0 (mod m).
Proof: We make use of identity 2.1, namely Fn+j = Lj Fn + (1)j+1 Fnj , in
the first line below.
Fnj + Fn Fn+j
Fnj + Fn Lj Fn + (1)j+1 Fnj
Fn Lj Fn (since j is odd)
(Lj 1)Fn 0
(mod m).
We are now ready to present the main theorem of this section.

Theorem 4.11 If Fj F2j (mod m), then {Fn , Fn+j } is a Fibonacci subsequence for all
n such that Fn 0 (mod m).
Proof:
We will look at two cases: j is odd and j is even. In each case we
will show that {F0 , Fj } must be a Fibonacci subsequence, then application of

lemma 4.9 will imply the conclusion of the theorem.
Case 1: Suppose j is odd.
In lemma 4.10 we can view (Lj 1)Fn 0 (mod m) as a linear congruence
with Fn given and (Lj 1) the variable. The solution set of this linear congruence
(dn 1)m
3m
} where dn = gcd(Fn , m). If (Lj 1) is
is (Lj 1) {0, dmn , 2m
dn , dn , . . . ,
dn
congruent to any of the elements in the set, then the linear congruence will be
satisfied. [4, p. 78].
Let n be given. Then
{odd j : Fnj + Fn Fn+j }

m 2m 3m
(dn 1)m
=
odd j : (Lj 1) 0, ,
,
, . . . , or
.
dn dn dn
dn
We know that Fn |Ftn so we must have gcd(Fn , m)| gcd(Ftn , m). Thus, using
(dn 1)m
} {0, dm
, 2m , . . . , (dtnd1)m
}.
previous notation, dn |dtn . It follows that {0, dmn , 2m
dn , . . . ,
dn
tn dtn
tn
58
Hence, if j is a solution given n, then j is a solution given any multiple of n. In

other words, if Fnj + Fn Fn+j (mod m) then Ftnj + Ftn Ftn+j (mod m)
for all t.
In particular, suppose that j is a solution given n = j (this is exactly what
the hypothesis of our theorem supposes). Then F0 + Fj F2j , Fj + F2j F3j ,
F2j + F3j F4j , . . . (mod m). That is, {F0 , Fj } is a Fibonacci subsequence.
Case 2: Suppose j is even.
First we look at a necessary condition. If we assume that {F0 , Fj } is, in fact,
a Fibonacci subsequence then we must have Fj F0 + Fj Fj (mod m).
However, we know from identity 1.3 that for even j, Fj = Fj , and so we
must conclude that Fj Fj (mod m). Thus if m is odd we must have Fj
0 (mod m), and if m is even we have either Fj 0 or
only nontrivial case occurs when m is even and Fj
m
2
m
2
(mod m). Hence, the
(mod m).
Once again, identity 2.1 states that Fn+j = Lj Fn + (1)j+1 Fnj . When we
let n = tj j we get Ftj = Lj F(t1)j + (1)j+1 F(t2)j . When t = 2 we get the
familiar identity F2j = Lj Fj , which implies here,
m
2
Lj ( m
2 ) (mod m). We use
this relation to simplify the congruences (taken mod m) below.

F3j
Lj F2j Fj
Lj (
F4j
m
m
)( )0
2
2
Lj F3j F2j
Lj (0) (
F5j
m
m
)
2
2
Lj F4j F3j
Lj (
m
m
) (0)
2
2
..
.
m
m m
Hence our subsequence {F0 , Fj } = 0, m
2 , 2 , 0, 2 , 2 , . . . will continue on in
this manner and {F0 , Fj } is a Fibonacci subsequence.

Now it is a simple matter to apply lemma 4.9 and see that if Fn 0 (mod m)
then {Fn , Fn+j } must also be a Fibonacci subsequence.
59
This result gives us a fairly easy method for finding many of the Fibonacci subsequences
of F (mod m), if any exist. However, this method may actually give us more than that as
the following conjecture asserts.
Conjecture 4.12 If 56 | m then every Fibonacci subsequence contains a zero.
If the above conjecture is true, then our method should give us all the Fibonacci subsequences of F (mod m) when 56 | m. It also appears that every time 5|m there must be at
least one Fibonacci subsequence containing no zeros. This may have something to do with
the fact that the period of the Lucas numbers = k(m) when 56 | m, and 15 k(m) when 5|m.
We also make the next conjecture, which tests for the existence of a Fibonacci subsequence when we are given four terms in a row and none of them are required to be congruent
to zero modulo m.
Conjecture 4.13 If Fnj + Fn Fn+j and Fn + Fn+j Fn+2j (mod m) then {Fn , Fn+j }
is a Fibonacci subsequence.
Finding just three consecutive terms of a subsequence which exhibit the Fibonacci recurrence relation is not sufficient to conclude that the subsequence is a Fibonacci subsequence.
As a counter example we see that F7 , F9 , F11 are 1, 4, and 5 (mod 6) respectively, but
F13 5 (mod 6).
One thing that makes this area of research attractive is that these subsequences appear
quite frequently. Suppose we dont count rotations as different subsequences, and we dont
count trivial subsequences of all zeros or the Fibonacci sequence itself. Then for example
we find that F (mod 5) contains 8 Fibonacci subsequences, F (mod 6) contains 11 Fibonacci
subsequences, F (mod 7) contains 1 Fibonacci subsequence, and F (mod 10) contains 39 Fibonacci subsequences. Table 4.2 lists all the smallest pairs (n, n + j) for which {Fn , Fn+j }
is a Fibonacci subsequence modulo 10. Note that k(10) = 60 and (10) = 4.
Finally we mention that not only is the existence of these special subsequences interesting, but how they actually look has its intrigue. We have seen that if a Fibonacci subsequence exits with even spread, then it has the form ( m
2 )[F (mod 2)]. Every third residue
from F (mod 6) forms a sequence like 2[F (mod 3)]. Every seventh residue from F (mod 91)
forms a sequence like F (mod 7) multiplied through by 13.
60
j
1
5
9
10
13
15
17
20
21
25
29
30
33
35
37
40
41
45
49
50
53
55
57
60
gcd(j, 60)
1
5
3
10
1
15
1
20
3
5
1
30
3
5
1
20
1
15
1
10
1
5
3
60
(n, n + j)
(0,1)
(0,5) (1,6) (2,7) (3,8) (4,9)
(0,9)
(0,10) (5,15)
(0,13)
(0,15)*
(0,17)
(0,20) (5,25) (10,30) (15,45)
(0,21)
(0,25) (1,26) (2,27) (3,28) (4,29)
(0,29)
(0,30)* (15,45)*
(0,33)
(0,35)
(0,37)
(0,40) (5,45) (10,50) (15,55)
(0,41)
(0,45)* (3,48) (6,51) (9,54) (12,57)
(0,49)
(0,50) (5,55)
(0,53)
(0,55)
(0,57)
(0,60)* (15,75)* (30,90)* (45,105)*
Table 4.2: Subsequences of F (mod 10). The subsequences with asterisks are trivial subsequences, containing only zeros.
61
Once again, it appears that for every new insight we gain into the Fibonacci sequence,
a multitude of new relationships emerge to amaze and intrigue us.
62
Appendix A
The First 30 Fibonacci and

Lucas Numbers
n
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Fn
1
1
2
3
5
8
13
21
34
55
89
144
233
377
610
987
1597
2584
4181
6765
10946
17711
28657
46368
75025
121393
196418
317811
514229
832040
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
Ln
1
3
4
7
11
23
18
29
37
47
2 17
76
5 11
123
199
24 32
322
521
13 29
843
2 5 61
1364
3 7 47
2207
3571
23 17 19
5778
37 113
9349
3 5 11 41
15127
2 13 421
24476
89 199
39603
64079
25 32 7 23
103682
52 3001
167761
233 521
271443
2 17 53 109
439204
3 13 29 281
710647
1149851
23 5 11 31 61 1860498
63
22
2 32
= 22 19
= 3 41
= 2 7 23
= 3 281
= 22 11 31
= 2 33 107
=
=
=
=
=
=
=
=
=
=
=
7 2161
22 29 211
3 43 307
139 461
2 47 1103
11 101 161
3 90481
22 19 5779
72 14503
59 19489
2 32 41 2521
64
APPENDIX A. THE FIRST 30 FIBONACCI AND LUCAS NUMBERS
65
APPENDIX B. K(M ), (M ), AND (M ) FOR 2 M 1000
66
Appendix B
k(m), (m), and (m) for 2 m

1000
m
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
k(m)
3
8
6
20
24
16
12
24
60
10
24
28
48
40
24
36
24
18
60
16
30
48
24
100
84
72
48
14
120
30
48
40
36
80
24
(m)
3
4
6
5
12
8
6
12
15
10
12
7
24
20
12
9
12
18
30
8
30
24
12
25
21
36
24
14
60
30
24
20
9
40
12
(m)
1
2
1
4
2
2
2
2
4
1
2
4
2
2
2
4
2
1
2
2
1
2
2
4
4
2
2
1
2
1
2
2
4
2
2
m
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
k(m)
76
18
56
60
40
48
88
30
120
48
32
24
112
300
72
84
108
72
20
48
72
42
58
120
60
30
48
96
140
120
136
36
48
240
70
(m)
19
18
28
30
20
24
44
30
60
24
16
12
56
75
36
42
27
36
10
24
36
42
58
60
15
30
24
48
35
60
68
18
24
120
70
(m)
4
1
2
2
2
2
2
1
2
2
2
2
2
4
2
2
4
2
2
2
2
1
1
2
4
1
2
2
4
2
2
2
2
2
1
m
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
k(m)
24
148
228
200
18
80
168
78
120
216
120
168
48
180
264
56
60
44
120
112
48
120
96
180
48
196
336
120
300
50
72
208
84
80
108
(m)
12
37
57
100
18
40
84
78
60
108
60
84
24
45
132
28
30
11
60
56
24
60
48
90
24
49
168
60
150
50
36
104
42
40
27
(m)
2
4
4
2
1
2
2
1
2
2
2
2
2
4
2
2
2
4
2
2
2
2
2
2
2
4
2
2
2
1
2
2
2
2
4
67
m
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
k(m)
72
72
108
60
152
48
76
72
240
42
168
174
144
120
110
60
40
30
500
48
256
192
88
420
130
120
144
408
360
36
276
48
46
240
32
210
140
24
140
(m)
36
36
27
30
76
24
19
36
120
42
84
174
72
60
110
15
20
30
125
24
128
96
44
105
130
60
72
204
180
18
69
24
46
120
16
210
70
12
70
(m)
2
2
4
2
2
2
4
2
2
1
2
1
2
2
1
4
2
1
4
2
2
2
2
4
1
2
2
2
2
2
4
2
1
2
2
1
2
2
2
m
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
k(m)
444
112
228
148
600
50
36
72
240
60
168
316
78
216
240
48
216
328
120
40
168
336
48
364
180
72
264
348
168
400
120
232
132
178
120
90
336
120
48
(m)
111
56
114
37
300
50
18
36
120
30
84
79
78
108
120
24
108
164
60
20
84
168
24
91
45
36
132
87
84
200
60
116
33
178
60
90
168
60
24
(m)
4
2
2
4
2
1
2
2
2
2
2
4
1
2
2
2
2
2
2
2
2
2
2
4
4
2
2
4
2
2
2
2
4
1
2
1
2
2
2
m
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
k(m)
380
120
180
96
144
180
190
96
388
588
280
336
396
120
22
300
136
150
112
72
40
624
48
168
90
240
42
108
280
72
440
72
240
108
296
60
252
456
448
(m)
95
60
90
48
72
90
190
48
97
147
140
168
99
60
22
150
68
150
56
36
20
312
24
84
90
120
42
54
140
36
220
36
120
27
148
30
63
228
224
(m)
4
2
2
2
2
2
1
2
4
4
2
2
4
2
1
2
2
1
2
2
2
2
2
2
1
2
1
2
2
2
2
2
2
4
2
2
4
2
2
68
m
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
k(m)
48
600
228
456
72
114
240
80
84
52
168
160
174
312
144
238
120
240
330
648
60
560
120
252
60
168
1500
250
48
240
768
360
384
516
264
304
420
168
390
(m)
24
300
57
228
36
114
120
40
42
13
84
80
174
156
72
238
60
120
330
324
30
280
60
126
30
84
375
250
24
120
384
180
192
129
132
152
210
84
390
(m)
2
2
4
2
2
1
2
2
2
4
2
2
1
2
2
1
2
2
1
2
2
2
2
2
2
2
4
1
2
2
2
2
2
4
2
2
2
2
1
m
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
k(m)
176
120
540
144
88
408
268
360
270
72
112
276
100
48
556
138
120
240
56
96
568
210
360
420
80
48
612
420
392
444
588
336
580
228
360
444
336
600
176
(m)
88
60
135
72
44
204
67
180
270
36
56
69
50
24
139
138
60
120
28
48
284
210
180
210
40
24
153
210
196
222
147
168
290
114
180
111
168
300
88
(m)
2
2
4
2
2
2
4
2
1
2
2
4
2
2
4
1
2
2
2
2
2
1
2
2
2
2
4
2
2
2
4
2
2
2
2
4
2
2
2
m
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
k(m)
150
200
72
60
72
88
240
208
60
310
168
628
948
240
78
636
216
70
480
72
48
36
216
700
984
216
120
32
120
110
168
456
336
680
48
676
1092
152
180
(m)
150
100
36
15
36
44
120
104
30
310
84
157
237
120
78
159
108
70
240
36
24
18
108
175
492
108
60
16
60
110
84
228
168
340
24
169
273
76
90
(m)
1
2
2
4
2
2
2
2
2
1
2
4
4
2
1
4
2
1
2
2
2
2
2
4
2
2
2
2
2
1
2
2
2
2
2
4
4
2
2
69
m
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
k(m)
30
72
784
264
240
348
232
168
174
1200
504
240
236
696
140
132
144
534
358
120
342
90
440
336
740
120
736
48
120
1140
432
120
748
180
1000
96
28
144
378
(m)
30
36
392
132
120
87
116
84
174
600
252
120
59
348
70
66
72
534
358
60
342
90
220
168
185
60
368
24
60
285
216
60
187
90
500
48
14
72
378
(m)
1
2
2
2
2
4
2
2
1
2
2
2
4
2
2
2
2
1
1
2
1
1
2
2
4
2
2
2
2
4
2
2
4
2
2
2
2
2
1
m
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
k(m)
180
256
570
768
192
80
1164
264
588
388
840
144
336
520
396
780
120
796
66
144
600
200
408
420
150
1080
336
380
72
408
120
552
624
464
48
840
336
184
90
(m)
90
128
570
384
96
40
291
132
294
97
420
72
168
260
99
390
60
199
66
72
300
100
204
210
150
540
168
190
36
204
60
276
312
232
24
420
168
92
90
(m)
2
2
1
2
2
2
4
2
2
4
2
2
2
2
4
2
2
4
1
2
2
2
2
2
1
2
2
2
2
2
2
2
2
2
2
2
2
2
1
m
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
k(m)
418
240
84
42
96
108
900
840
240
72
280
1320
430
72
868
240
280
108
144
888
438
60
336
252
888
456
220
1344
296
96
448
600
40
228
200
456
560
72
916
(m)
418
120
21
42
48
54
225
420
120
36
140
660
430
36
217
120
140
54
72
444
438
30
168
63
444
228
55
672
148
48
224
300
20
114
100
228
280
36
229
(m)
1
2
4
1
2
2
4
2
2
2
2
2
1
2
4
2
2
2
2
2
1
2
2
4
2
2
4
2
2
2
2
2
2
2
2
2
2
2
4
70
m
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
k(m)
114
72
240
46
240
928
168
120
156
936
168
272
480
632
348
440
312
900
144
216
714
478
240
532
240
48
330
980
648
976
60
328
1680
490
120
252
252
120
120
(m)
114
36
120
46
120
464
84
60
39
468
84
136
240
316
174
220
156
450
72
108
714
478
120
133
120
24
330
245
324
488
30
164
840
490
60
126
126
60
60
(m)
1
2
2
1
2
2
2
2
4
2
2
2
2
2
2
2
2
2
2
2
1
1
2
4
2
2
1
4
2
2
2
2
2
1
2
2
2
2
2
m
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
k(m)
560
168
498
1500
336
750
1008
48
100
240
728
768
254
360
592
768
72
516
1040
264
160
912
696
420
26
168
1048
390
400
528
180
120
1104
540
696
144
280
264
360
(m)
280
84
498
750
168
750
504
24
50
120
364
384
254
180
296
384
36
129
520
132
80
456
348
210
26
84
524
390
200
264
90
60
552
135
348
72
140
132
180
(m)
2
2
1
2
2
1
2
2
2
2
2
2
1
2
2
2
2
4
2
2
2
2
2
2
1
2
2
1
2
2
2
2
2
4
2
2
2
2
2
m
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
k(m)
408
712
804
560
360
90
270
360
144
540
336
1096
276
120
300
126
48
624
1668
760
138
124
120
616
240
360
168
376
96
380
1704
432
420
568
360
570
420
760
240
(m)
204
356
201
280
180
90
270
180
72
135
168
548
138
60
150
126
24
312
417
380
138
31
60
308
120
180
84
188
48
95
852
216
210
284
180
570
210
380
120
(m)
2
2
4
2
2
1
1
2
2
4
2
2
2
2
2
1
2
2
4
2
1
4
2
2
2
2
2
2
2
4
2
2
2
2
2
1
2
2
2
71
m
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
k(m)
1200
96
1156
612
776
420
336
1176
540
444
840
588
1176
336
90
1740
792
456
1188
360
720
444
88
336
598
600
600
528
408
150
220
600
1216
144
112
60
224
72
1228
(m)
600
48
289
153
388
210
168
588
270
222
420
147
588
168
90
870
396
228
297
180
360
222
44
168
598
300
300
264
204
150
110
300
608
72
56
15
112
36
307
(m)
2
2
4
4
2
2
2
2
2
2
2
4
2
2
1
2
2
2
4
2
2
2
2
2
1
2
2
2
2
1
2
2
2
2
2
4
2
2
4
m
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
k(m)
264
40
240
1236
624
206
60
144
930
176
168
2500
1884
360
948
684
240
630
156
168
636
1280
216
112
210
840
960
640
72
1288
48
440
36
1296
216
290
2100
240
984
(m)
132
20
120
309
312
206
30
72
930
88
84
625
471
180
474
171
120
630
78
84
159
640
108
56
210
420
480
320
36
644
24
220
18
648
108
290
525
120
492
(m)
2
2
2
4
2
1
2
2
1
2
2
4
4
2
2
4
2
1
2
2
4
2
2
2
1
2
2
2
2
2
2
2
2
2
2
1
4
2
2
m
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
k(m)
1308
216
260
120
888
96
658
120
220
330
504
168
720
456
336
336
448
2040
60
48
1348
2028
1800
1092
452
456
784
180
456
30
1368
72
1380
2352
456
264
756
240
138
(m)
327
108
130
60
444
48
658
60
55
330
252
84
360
228
168
168
224
1020
30
24
337
507
900
546
113
228
392
90
228
30
684
36
345
1176
228
132
189
120
138
(m)
4
2
2
2
2
2
1
2
4
1
2
2
2
2
2
2
2
2
2
2
4
4
2
2
4
2
2
2
2
1
2
2
4
2
2
2
4
2
1
72
m
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
k(m)
348
240
696
460
168
360
174
104
1200
700
504
684
480
160
708
400
696
118
420
312
132
240
144
140
534
952
1074
718
120
208
342
240
90
700
1320
1456
336
1944
2220
(m)
174
120
348
230
84
180
174
52
600
175
252
342
240
80
177
200
348
118
210
156
66
120
72
70
534
476
1074
718
60
104
342
120
90
350
660
728
168
972
555
(m)
2
2
2
2
2
2
1
2
2
4
2
2
2
2
4
2
2
1
2
2
2
2
2
2
1
2
1
1
2
2
1
2
1
2
2
2
2
2
4
m
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
k(m)
792
120
1468
2208
560
48
680
120
738
1140
504
432
496
120
740
2244
168
180
144
3000
750
96
1000
84
100
144
1516
378
240
180
380
768
432
570
360
768
812
384
192
(m)
396
60
367
1104
280
24
340
60
738
570
252
216
248
60
185
561
84
90
72
1500
750
48
500
42
50
72
379
378
120
90
95
384
216
570
180
384
406
192
96
(m)
2
2
4
2
2
2
2
2
1
2
2
2
2
2
4
4
2
2
2
2
1
2
2
2
2
2
4
1
2
2
4
2
2
1
2
2
2
2
2
m
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
k(m)
240
1032
1164
1548
264
300
588
304
1164
360
840
70
144
504
336
1580
1560
1576
396
176
780
304
120
420
2388
1080
66
228
144
288
1200
264
600
740
408
240
420
536
300
(m)
120
516
582
387
132
150
294
152
291
180
420
70
72
252
168
395
780
788
198
88
390
152
60
105
597
540
66
57
72
144
600
132
300
370
204
120
210
268
150
(m)
2
2
2
4
2
2
2
2
4
2
2
1
2
2
2
4
2
2
2
2
2
2
2
4
4
2
1
4
2
2
2
2
2
2
2
2
2
2
2
73
m
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
k(m)
202
1080
270
336
1080
1140
1640
72
792
408
336
120
820
552
1648
624
200
1392
1656
48
276
840
1112
672
1008
552
1680
90
360
1254
838
240
406
84
56
42
1820
96
880
(m)
202
540
270
168
540
570
820
36
396
204
168
60
205
276
824
312
100
696
828
24
69
420
556
336
504
276
840
90
180
1254
838
120
406
21
28
42
455
48
440
(m)
1
2
1
2
2
2
2
2
2
2
2
2
4
2
2
2
2
2
2
2
4
2
2
2
2
2
2
1
2
1
1
2
1
4
2
1
4
2
2
m
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
k(m)
216
568
900
912
840
1708
240
360
72
1716
840
78
1320
80
1290
1728
144
1740
2604
1224
240
390
840
952
108
1176
144
2000
888
1756
438
1176
120
176
336
1768
252
1160
888
(m)
108
284
225
456
420
427
120
180
36
429
420
78
660
40
1290
864
72
435
651
612
120
390
420
476
54
588
72
1000
444
439
438
588
60
88
168
884
126
580
444
(m)
2
2
4
2
2
4
2
2
2
4
2
1
2
2
1
2
2
4
4
2
2
1
2
2
2
2
2
2
2
4
1
2
2
2
2
2
2
2
2
m
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
k(m)
1776
456
256
660
1080
1344
288
888
1780
192
336
1344
210
600
108
120
176
228
180
600
1816
456
600
1680
70
72
840
2748
120
114
1040
72
102
240
88
138
140
240
1900
(m)
888
228
128
165
540
672
144
444
890
96
168
672
210
300
27
60
88
114
90
300
908
228
300
840
70
36
420
687
60
114
520
36
102
120
44
138
70
120
475
(m)
2
2
2
4
2
2
2
2
2
2
2
2
1
2
4
2
2
2
2
2
2
2
2
2
1
2
2
4
2
1
2
2
1
2
2
1
2
2
4
74
m
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
k(m)
2784
624
336
928
120
1008
156
1240
936
180
168
1876
816
1256
480
470
1896
240
696
720
1320
1896
312
1036
900
(m)
1392
312
168
464
60
504
78
620
468
90
84
469
408
628
240
470
948
120
348
360
660
948
156
259
450
(m)
2
2
2
2
2
2
2
2
2
2
2
4
2
2
2
1
2
2
2
2
2
2
2
4
2
m
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
k(m)
1272
144
212
216
380
714
280
1434
1104
480
930
1596
72
240
1940
48
176
660
72
2940
970
648
368
2928
1400
(m)
636
72
53
108
190
714
140
1434
552
240
930
399
36
120
485
24
88
330
36
735
970
324
184
1464
700
(m)
2
2
4
2
2
1
2
1
2
2
1
4
2
2
4
2
2
2
2
4
1
2
2
2
2
m
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
k(m)
120
652
984
220
1680
216
1470
1968
120
1980
252
32
252
528
120
198
240
440
1680
220
168
1996
498
1368
1500
(m)
60
163
492
110
840
108
1470
984
60
495
126
16
126
264
60
198
120
220
840
110
84
499
498
684
750
(m)
2
4
2
2
2
2
1
2
2
4
2
2
2
2
2
1
2
2
2
2
2
4
1
2
2
75
APPENDIX C. ONE PERIOD OF F (MOD M ) FOR 2 M 50
76
Appendix C
One Period Of F (mod m) for 2

m 50
m
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
0
0
0
0
0
3
0
0
0
6
0
5
5
0
0
9
0
5
0
3
7
0
0
0
13
0
16
0
15
0
0
5
15
0
0
11
0
3
2
0
21
1
1
1
1
1
2
1
1
1
2
1
6
1
1
1
2
1
0
1
12
13
1
11
1
2
1
15
1
2
1
1
6
1
1
1
12
1
21
13
1
2
1
1
1
1
1
5
1
1
1
8
1
1
6
1
1
11
1
5
1
1
6
1
11
1
15
1
14
1
17
1
1
11
16
1
1
1
1
1
15
1
23
2
2
2
2
1
2
2
2
1
2
7
7
2
2
1
2
5
2
13
5
2
7
2
1
2
12
2
1
2
2
17
17
2
2
13
2
22
5
2
1
0
3
3
3
2
1
0
5
3
2
3
1
1
3
4
4
0
1
4
5
4
0
3
5
2
5
0
4
3
3
3
5
5
5
1
0
8
6
5
4
0
5
3
6
2
7
6
7
1
5
1
8
3
8
3
3
3
5
5
0
5
5
8
3
3
8
8
3
8
3
2
1
1
1
6
10
9
4
9
9
1
10
5
0
5
9
9
4
4
9
9
3
8
3
3
10
3
0
11
3
3
3
5
2
5
13
2
5
10
5
8
12
8
13
13
8
13
8
0
1
13
12
1
13
8
13
11
7
11
6
9
13
6
6
6
5
4
14
2
3
9
3
5
4
5
8
13
8
13
0
13
4
13
3
3
3
8
13
3
3
14
3
0
20
3
5
5
5
10
5
5
5
5
22
2
5
8
8
13
3
8
8
19
8
22
22
8
13
13
18
13
13
13
2
13
21
1
13
2
3
2
1
4
4
1
5
7
7
2
0
5
5
7
2
7
7
7
2
4
9
9
1
6
1
10
12
12
12
11
10
5
1
4
7
9
8
13
1
8
9
7
10
1
5
8
1
9
6
10
5
7
14
4
9
9
9
0
8
13
9
2
7
9
10
5
2
12
12
11
7
2
13
4
14
8
11
1
5
0
13
16
4
9
1
4
5
17
8
14
0
12
2
17
3
16
17
15
1
16
16
16
15
13
10
2
1
11
16
0
21
21
21
20
15
14
9
9
13
12
1
11
18
17
15
0
5
13
11
13
9
9
14
5
1
11
4
9
19
18
12
5
13
18
13
2
13
16
17
7
12
20
3
2
10
5
5
1
16
18
7
12
17
1
17
17
2
4
9
19
1
6
1
19
13
10
9
15
20
10
6
2
3
12
9
14
12
3
21
17
10
20
8
14
18
11
21
10
17
17
17
10
13
16
77
m
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
0
15
5
20
10
0
5
1
16
23
0
15
24
18
0
17
7
0
0
15
15
0
15
15
0
7
0
13
11
0
0
0
33
0
10
0
25
0
33
0
31
36
12
0
0
18
27
0
5
35
1
21
16
11
6
1
0
25
5
2
1
11
22
8
1
26
13
1
1
26
1
11
16
11
1
3
1
2
13
1
23
1
32
1
26
6
16
1
2
1
31
35
18
1
1
26
25
1
26
21
1
11
21
6
16
1
5
0
21
25
1
26
19
26
1
15
20
1
1
11
16
11
1
26
1
10
1
15
24
1
23
1
31
1
1
6
6
1
35
1
25
34
30
1
1
5
13
1
31
16
2
7
12
17
22
2
5
25
0
1
2
10
14
7
2
13
5
2
2
7
17
22
17
7
2
13
2
17
5
2
13
2
29
2
27
12
22
2
1
2
19
32
11
2
2
31
38
2
17
37
3
18
8
23
13
3
10
25
21
5
0
20
15
10
5
15
24
21
8
18
3
13
23
8
25
23
16
13
18
23
3
8
13
14
21
11
21
11
1
16
6
21
13
18
1
9
4
24
19
14
8
1
13
12
5
15
0
10
20
3
14
5
13
14
19
24
4
9
11
15
18
25
19
9
24
14
4
14
3
23
12
8
3
23
18
13
25
18
15
11
2
12
22
7
17
13
21
12
23
10
15
20
0
5
12
13
1
8
12
2
17
7
22
25
8
13
5
22
17
12
7
2
11
21
14
13
9
19
4
14
24
10
3
1
18
6
11
16
21
1
21
24
15
5
3
9
6
6
3
0
25
3
3
18
3
3
18
3
3
23
3
0
29
3
3
3
26
3
28
18
28
3
5
19
20
13
5
13
2
5
5
25
20
25
5
10
5
5
5
17
2
5
16
5
21
5
20
30
15
5
8
1
26
19
8
13
27
8
8
13
23
28
23
13
8
28
8
17
31
8
19
8
13
8
13
13
8
8
13
20
19
5
13
26
1
13
13
8
13
23
28
23
13
2
13
2
1
13
2
13
0
13
33
8
23
13
21
21
18
24
21
11
7
14
10
2
6
9
1
8
1
26
27
20
8
22
11
1
5
1
9
3
12
17
25
23
26
1
8
16
26
4
15
0
12
4
26
16
19
26
1
23
25
17
4
21
9
22
13
15
22
9
7
24
1
5
8
1
9
6
21
21
21
6
21
21
6
21
30
21
19
5
4
29
19
14
19
29
3
1
2
21
26
25
20
25
5
10
5
24
2
29
19
14
19
29
4
27
28
24
9
9
24
9
9
20
1
23
28
23
13
8
13
16
17
7
2
7
17
22
5
10
5
25
20
25
5
21
27
12
27
27
12
27
26
7
17
22
17
7
2
16
4
29
19
14
19
29
11
11
16
11
1
26
1
27
23
8
25
29
16
5
9
2
25
7
2
9
27
16
29
25
24
9
21
2
21
21
21
13
21
11
21
31
21
1
23
0
13
34
9
29
19
34
22
11
21
26
20
20
15
15
19
23
1
21
5
19
29
9
34
17
12
12
8
31
4
14
24
14
0
2
13
29
2
23
8
33
13
17
14
25
3
33
27
22
22
27
17
16
5
32
1
15
30
20
5
34
30
30
1
13
2
33
10
32
0
23
1
33
7
17
7
32
15
22
12
27
2
13
29
29
34
34
28
16
6
26
1
5
3
7
29
4
3
3
36
12
3
8
13
5
26
24
15
5
5
28
11
5
25
10
8
33
16
19
8
8
25
23
8
33
23
13
22
3
34
13
13
14
34
13
18
33
21
18
19
16
21
21
0
18
21
11
16
34
3
22
13
34
34
14
13
34
29
9
18
21
4
29
17
16
14
31
15
0
25
15
24
26
5
13
11
28
5
9
29
34
33
8
30
34
30
27
3
36
24
29
19
11
32
19
2
5
38
31
2
33
18
13
7
3
12
36
35
26
34
38
17
7
32
18
35
31
1
2
25
26
1
10
25
5
25
1
6
6
36
0
31
0
6
0
36
6
37
12
21
1
37
8
10
29
8
37
27
32
37
37
17
2
24
9
39
21
26
1
APPENDIX C. ONE PERIOD OF F (MOD M ) FOR 2 M 50
78
m
41
42
43
44
45
46
47
48
49
50
0
0
0
3
21
0
14
3
2
22
0
33
0
15
15
0
30
30
0
3
25
0
44
0
45
0
3
7
46
35
3
0
15
5
20
35
25
40
5
45
10
25
15
30
45
35
1
40
1
26
13
1
24
41
37
13
1
34
1
11
31
26
16
41
1
44
13
1
42
1
2
1
19
27
44
43
40
1
46
41
11
6
1
21
16
11
31
26
21
41
36
31
1
40
1
29
34
1
38
1
39
35
1
23
1
26
1
26
1
26
1
1
38
1
39
1
47
1
22
34
41
29
43
1
11
46
31
41
26
11
21
6
41
1
36
21
31
16
2
39
2
13
5
2
19
42
33
5
2
13
2
37
32
7
17
22
2
45
5
2
34
2
1
2
41
12
36
23
34
2
7
37
42
47
27
32
37
17
22
27
7
12
17
47
3
38
3
0
39
3
14
0
29
40
3
36
3
18
33
33
18
3
3
0
43
3
26
3
5
36
5
13
2
5
33
42
19
2
5
5
5
10
20
40
35
25
5
45
2
5
13
5
8
33
8
13
41
8
4
42
5
42
8
41
8
28
8
28
8
28
8
45
45
8
39
8
13
28
13
26
1
13
37
41
24
1
13
2
13
38
28
23
43
8
13
44
1
13
5
13
3
14
46
28
3
28
3
18
33
23
38
3
43
8
23
13
28
43
33
48
13
5
6
9
15
26
13
5
25
20
15
35
30
25
45
40
35
5
0
45
15
10
8
20
6
43
29
41
8
43
3
38
23
33
18
3
13
48
33
43
28
13
23
13
26
15
9
6
5
13
18
23
3
8
13
43
48
3
33
38
43
23
28
33
21
20
21
39
34
7
34
23
14
27
13
20
7
34
5
1
21
20
18
21
28
13
23
22
8
33
41
1
36
5
22
23
3
38
21
24
39
2
1
5
1
40
22
29
40
1
23
34
21
41
40
29
34
35
38
10
12
33
35
39
3
25
30
6
15
15
22
2
18
40
9
8
33
12
31
10
8
9
40
18
41
21
28
28
6
30
25
3
4
8
10
31
10
38
35
34
21
43
21
21
36
6
6
36
21
43
34
1
34
14
19
29
4
44
34
41
11
12
13
25
38
19
13
32
10
35
10
35
10
35
9
38
44
4
29
19
14
34
43
33
9
39
39
9
24
24
6
25
8
43
23
28
38
13
3
12
17
37
17
37
17
37
9
37
25
35
40
20
10
5
12
3
42
27
12
12
27
42
21
40
22
17
7
32
37
2
33
43
19
44
19
44
19
44
8
37
41
16
26
31
11
1
41
34
21
44
21
34
2
34
8
46
7
42
1
41
45
46
46
46
45
41
41
34
27
13
40
21
46
21
3
35
46
21
11
26
41
31
46
11
1
16
31
21
36
1
41
6
34
23
36
12
41
2
34
29
49
44
39
9
4
49
19
14
9
29
24
19
39
6
20
8
15
27
48
5
40
25
35
20
5
15
0
35
45
30
15
25
10
45
40
43
44
27
19
1
39
19
24
29
9
14
19
49
4
9
39
44
49
29
34
46
14
3
42
46
37
8
47
20
16
34
22
1
13
13
22
30
48
33
29
7
3
0
46
42
29
33
48
30
22
36
36
48
27
15
16
20
47
8
37
44
9
49
14
29
19
34
49
39
4
19
9
24
39
29
33
28
23
43
38
33
3
48
43
13
8
3
23
18
13
27
37
22
7
17
2
37
47
32
17
27
12
47
7
42
10
15
45
0
5
35
40
45
25
30
35
15
20
25
5
37
2
17
7
22
37
27
42
7
47
12
27
17
32
47
47
17
12
7
27
22
17
37
32
27
47
42
37
7
2
34
19
29
14
49
9
44
29
39
24
9
19
4
39
49
31
36
41
21
26
31
11
16
21
1
6
11
41
46
1
Bibliography
[1] Brother U. Alfred, Primes which are Factors of All Fibonacci Sequences,
Fibonacci Quarterly, 2 (1964), pp. 33-38.
[2] Agnes Andreassian, Fibonacci Sequences Modulo M, Fibonacci Quarterly, 12 (1974),
pp. 51-64.
[3] Huseyin Aydin and Geoff C. Smith, Fourier Analysis in Finite Nilpotent Groups,
Applications of Fibonacci Numbers, vol. 5 (1993), pp. 49-59.
[4] David M. Burton, Elementary Number Theory, Third Edition, Wm. C. Brown Publishers, Dubuque, Iowa, 1994.
[5] Paul A. Catlin, A Lower Bound for the Period of the Fibonacci Series Modulo M,
[6] Stuart Clary and Paul Hemenway, On Sums of Cubes of Fibonacci Numbers,
Applications of Fibonacci Numbers, vol. 5 (1993), pp. 123-136.
[7] Amos Ehrlich, On the Periods of the Fibonacci Sequence Modulo M,
[8] Herta T. Freitag, A Property of Unit Digits of
Fibonacci Numbers and Their Applications, 1984, pp. 39-41.
Fibonacci
Numbers,
[9] John H. Halton, On the Divisibility Properties of Fibonacci Numbers,

[10] E. T. Jacobson, Distribution of the Fibonacci Numbers Mod 2k , Fibonacci Quarterly,
30 (1992), pp. 211-215.
[11] Judy Kramer and Verner E. Hoggatt, Jr., Special Cases of Fibonacci Periodicity,
[12] Lawrence Kuipers and Jau-Shyong Shiue, A Distribution Property of the Sequence of
Fibonacci Numbers, Fibonacci Quarterly, 10 (1972), pp. 375-376, 392.
[13] S.
E.
Mamangakis,
Remarks on the Fibonacci Series Modulo m, American Mathematical Monthly, 68
(1961), pp. 648-649.
[14] Harald Niederreiter, Distribution fo the
79
Fibonacci
Numbers
Mod
5k ,
80
BIBLIOGRAPHY
[15] D. W. Robinson, The Fibonacci Matrix Modulo m, Fibonacci Quarterly, 1 (1963),

pp. 29-36.
[16] Ken Siler Fibonacci Summations, Fibonacci Quarterly, 1 (1963), pp. 67-70.
[17] T. E. Stanley, Some Remarks on the Periodicity of the Sequence of Fibonacci Numbers, Fibonacci Quarterly, 14 (1976), pp. 52-54.
[18] T. E. Stanley, Powers of the Period Function for the Sequence of Fibonacci Numbers,
[19] T. E. Stanley, Some Remarks on the Periodicity of the Sequence of Fibonacci Numbers
- II, Fibonacci Quarterly, 18 (1980), pp. 45-47.
[20] Steven Vajda, Fibonacci & Lucas Numbers, and the Golden Section. Ellis Horwood
Limited, Chichester, England, 1989.
[21] John Vinson, The Relation of the Period Modulo m to the Rand of Apparition of m
in the Fibonacci Sequence, Fibonacci Quarterly, 1 (1963), pp. 37-45.
[22] D. D. Wall, Fibonacci Series Modulo m, American Mathematical Monthly, 67 (1960),
pp. 525-532.

Fib Thesis

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fib Thesis

Uploaded by

Copyright:

Available Formats

The Fibonacci Sequence Under Various Moduli

3 Modular Representations of Fibonacci

B k(m), (m), and (m) for 2 m 1000

C One Period Of F (mod m) for 2 m 50

Leonardo of Pisa and the

CHAPTER 1. LEONARDO OF PISA AND THE FIBONACCI SEQUENCE

Thus the answer to Fibonaccis problem is 144.

Proof: Let m be fixed and we will proceed by inducting on n. When n = 1,

CHAPTER 1. LEONARDO OF PISA AND THE FIBONACCI SEQUENCE

= (Fn1 + Fn )Fn1 Fn2

(1)n (F1 F1 F02 )

CHAPTER 1. LEONARDO OF PISA AND THE FIBONACCI SEQUENCE

A generalized Fibonacci sequence, G, is one in which the usual recurrence relation

Identity 1.9 Ln = Fn1 + Fn+1 .

Proof: We will prove the identity by induction. It is easy to see that L1 = 1 =

Identity 1.10 F2n = Fn Ln .

Proof: F2n = Fn+n = Fn1 Fn + Fn Fn+1 = Fn (Fn1 + Fn+1 ) = Fn Ln .

We make a distinction between a Fibonacci sequence, meaning any generalized Fibonacci

Identity 1.11 Gm+n = Fn1 Gm + Fn Gm+1 .

which is true since F1 = 1, F0 = 0, and F1 = 1. By adding the two equations

CHAPTER 1. LEONARDO OF PISA AND THE FIBONACCI SEQUENCE

The Binet Formula

Note the following properties of and :

Now 1, , 2 , 3 , . . . and 1, , 2 , 3 , . . . are in fact generalized Fibonacci sequences since

As we see, ( n + n )+( n+1 + n+1 ) = ( n + n+1 )+( n + n+1 ) = n+2 + n+2 .

CHAPTER 2. THE BINET FORMULA

For example, in the Fibonacci sequence F0 = 0 and F1 = 1, and so = 1/ 5 and

Fi xi . It follows that g(x) F0 x0 F1 x1 = g(x) x. Hence

as the sums of geometric series we get

The coefficient of xn , in other words Fn , is ( n n )/ 5, just as we suspected.

CHAPTER 2. THE BINET FORMULA

Identity 2.1 Fm+n = Ln Fm + (1)n+1 Fmn .

m+n (1)n mn + (1)n mn m+n (1)n mn + (1)n mn

= [(1 + 5)n (1 5)n ]

CHAPTER 2. THE BINET FORMULA

Identity 2.4 Fn+1 = b Fn + 12 c for n 2.

CHAPTER 3. MODULAR REPRESENTATIONS OF FIBONACCI SEQUENCES

= (1)t (Fs Ft+1 Fs+1 Ft ).

3.1. THE PERIOD

Let h = h(m). We need to show that G(mod n) repeats in blocks

of length h. We do this by showing that Gi Gi+h (mod n) regardless of our

pei i . Then h(m) = lcm[h(pei i )],

CHAPTER 3. MODULAR REPRESENTATIONS OF FIBONACCI SEQUENCES

for all i. Since all the

are relatively prime, the Chinese Remainder Theo-

3.1. THE PERIOD

Proof: By inspection, k(2) = 3 and k(4) = 6, so the theorem holds for e = 1, 2.

Thus we know k(2r+1 )|3 2r .

CHAPTER 3. MODULAR REPRESENTATIONS OF FIBONACCI SEQUENCES

22r 0 (mod 2r+1 )

3.1. THE PERIOD

Proof: Since U k(p ) = I + pe B we can write U pk(p ) = (I + pe B)p = I p +

By assumption k(pe ) 6= k(pe+1 ) and so we know that U k(p ) = I + pe B where

CHAPTER 3. MODULAR REPRESENTATIONS OF FIBONACCI SEQUENCES

p = 2 in theorem 3.5 we may assume p 3.) That is, U k(p

I + pe+1 B 6 I (mod pe+2 ). This implies k(pe+1 ) 6= k(pe+2 ) so we must have

We will also use the following lemma:

Using the Legendre symbol and the law of quadratic reciprocity we

know that ( p5 ) = ( p5 ) since 5 is a prime of the form 4t + 1. Then,

3.1. THE PERIOD

and Fermats theorem we get Fp 5

(mod p). Now from the lemma we know that

1 (mod p) if and only if p = 5t 1, and 5