Binary Sequences Without Isolated Ones: Good A N

BINARY SEQUENCES WITHOUT ISOLATED ONES
Hi CHARD AUSTIN and RICHARD GUY

University of Calgary, Calgary, Canada
Liu [2] asks for the number of sequences of zeros and ones of length five, such that every digit 1 has at least
one neighboring 1. The solution [1] uses the principle of inclusion-exclusion, although it is easier in this particularcase to enumerate the twelve sequences:
00000, 11000,01100,00110,00011, 11100,01110,00111, 11110,01111, 11011, 11111.

In order to obtain a general result it seemed to us easier to find a recurrence relation.
Call a sequence good if each one in it has a neighboring one, and let an be the number of good sequences of
length n. For example,
a
I 32 = 2, a$ = 4, a4 = 7 and a^ = 12.
Good sequences of length n are obtained from other good sequences of length n - 1 by appending 0 or 1
to them, except that
(a) some not good sequences are also produced, namely those which end in 01, but are otherwise good, and
(b) there are good sequences which are not produced in this way; those obtained by appending 011 toa
good sequence of length n - 3.
So
(D
an = 2an,1 - an_2 + an-3
Alternatively, all good sequences are obtained from shorter good sequences by appending 0,11 or0111, so
that
(2)
an = ani +an_2 + an-4-
The characteristic equation for (2) is the same as that for (1), namely
x3 -2x2 +x- 1 = Of
(3)
except for the additional root - 1 . The equation (3) has one real root, y 1.754877666247 and two complex roots, a t'P, the square of whose modulus, 1/7, is less than 1.
an = cyn + (a+/b)(a*-/p)n + (a -
ih)(a-i$)n,
where
a = 1 -Y2y 0.122561166876, 0 = J2y*-4yl2

0.744861766619,
2
a = ( 7 - 2 T + 2 ) / 2 ( 2 T - 2 T + 3 ) 0.138937790848, b = ( 2 ? + 1 ) ( ? - D/2/3 - 0.202250124098,
2
c = ( y 2 + 1 ) / ( 2 7 2 - 2 7 + 3 ) 0.722124418303,
and an is the nearest integer to cyn.
The sequence {an } does not appear in Neil Sloane's book [ 3 ] ; nor do the corresponding sequences {afi*' }
of numbers of binary sequences of length n in which the ones occur only in blocks of length at least k. The
problem so farconsidered is k= 2. The more general analogs of (1), (2), (3) are
(1')
an = 2an-i - an2 + an_k-i,
(2')
(3')
an = ani+ ank + an-k~2 + an-k-3 + '" + an-2k >

xk+1-2xk+xk'11 = O.
Then
84
FEB. 1978
,(h)
=
-i
85
&) =
1 + 1Mr+1)(r + 2)
1;
k-i
for 0 < r < k; and for larger values of n, a ^ is the nearest integer to c^ y\, where y^ is the real root of (3')
which lies between 1 and 2, and c^ is an appropriate constant. Approximate values of y$, and cy, for k- 1(1)9
are shown in Table 1.
Table 1
a
k
yk
ck
1
2
1
2
1.7549
0.7221
1.6180
0.5854
1.5289
0.5033
1.4656
0.4481
1.4178
0.4082
1.3803
0.3778
1.3499
0.3539
1.3247
0.3344
The sequence {a } is similar to the Lucas sequence associated with the Fibonacci numbers, since 73 =
(1 + V5 )/2, the golden number.
The characteristic polynomial for (2') is the product of that for (V) with the cyclotomic polynomial x
+
k 2
+ ... + x + 1 \/\/nen /< j s odd, (3')js of even degree and is reducible and has a second real root between 0
x ~
and - 1 . Table 2 gives the values of ah" ' for/7 = 0(1)26, A: = 2(1)9. Of course, a^ = 2n, the number of unrestricted binary sequences of length n.
Table 2
*W
0
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
1
1
2
4
7
12
21
37
65
114
200
351
616
1081
1897
3329
5842
10252
17991
31572
55405
97229
170625
299426
525456
922111
1618192
,P)
1
1
1
2
4
7
11
17
27
44
72
117
189
305
493
798
1292
2091
3383
5473
8855
14328
23184
37513
60697
98209
158905
JV
J 5)
J 6)
1
1
1
1
2
4
7
11
16
23
34
52
81
126
194
296
450
685
1
1
1
1
1
2
4
7
11
16
22
30
42
61
91
137
205
303
443
644
936
1
1
1
1
1
1
2
4
7
11
16
22
29
38
51
71
102
149
218
316
452
639
897
1046
1601
2452
3753
5739
8771
13404
20489
31327
1365
1999
2936
4316
6340
9300
1257
1766
2493
3536
JV
2
4
7
11
16
22
29
37
47
61
82
114
162
232
331
467
650
894
1220
1660
JS) (9)
1
1
1
1
1
1
1
1
2
4
7
11
16
22
29
37
46
57
72
94
127
176
247
347
484
667
907
2
4
7
11
16
22
29
37
46
56
68
84
107
141
191
263
364
502
Since these are recurring sequences, they have many divisibility properties. Examples are 5 ^ ^ just if n =
- 4 or - 2 , mod 12; 8 | a ^ just if n = - 4 o r - 2 , mod 14 and 2|a^ ' according to the residue class to which n
belongs, mod 2(2(k+1)/2
- 1), k odd, or mod 2k+1
- 1, k even.
86
FEB. 1978
REFERENCES
1. Murray Edelberg, Solutions to Problems in 2, McGraw-Hill, 1968, p. 74.
2. C.LLiu, Introduction to Combinatorial Mathematics, McGraw-Hill, 1968, Problem 4 - 4 , p. 119.
3. N. J. A. Sloane, A Handbook of Integer Sequences, Academic Press, 1973, p. 59.
*******
ON THE EQUALITY OF PERIODS OF DIFFERENT
MODULI IN THE FIBONACCI SEQUENCE
JAMES E. DESMOND
Pensacola Junior College, Pensacola, Florida 32504
Let m be an arbitrary positive integer. According to the notation of Vinson [1, p. 37] let s(m) denote the period
of Fn modulo m and let f(m) denote the rank of apparition of m in the Fibonacci sequence.
Let p be an arbitrary prime. Wall [2, p. 528] makes the following remark: "The most perplexing problem we have
met in this study concerns the hypothesis s(p2) ? sip). We have run a test on a digital computer which shows that
s(p ) t s(p) for all p up to 10,000; however, we cannot yet prove that s(p2) = s(p) is impossible.The question is
closely related to another one, "can a number x have the same order mod p and mod p2?," for which rare cases give
an affirmative answer (e.g., x = 3, p = 11;x = 2,p = 1093); hence, one might conjecture that equality may hold for
some exceptional p."
Based on Ward's Last Theorem [3, p. 205] we shall give necessary and sufficient conditions iorsfp ) =s(p).
From Robinson [4, p. 30] we have for/77,/7 > 0
(1)
Fn+r = Fr (mod m) far all integers r if and only if s(m) \n.
If m,n > 0 and m \n, then Fsrnj+r = Fr (mod m) for all r. Therefore by (1), s(m) \s(n). So we have for m,n > O
(2)
m\n implies s(m)\s(n).
It is easily verified that for all integers n
(3)
F2n+1
(-Dn
+F
n+lLn-
From Theorem 1 of [1, p. 39] we have X\\dXs(m) is even if m > 2.

An equivalent form of the following theorem can be found in Vinson [1, p. 42].
Theorem 1. We have
i) s(m) = 4f(m) if and only if m > 2 and f(m) is odd.
ii) s(m) = f(m) if and only if m = 1 or 2 and s(m)/2 is odd.
iii) s(m) =2f(m) if and only if f(m) is even and s(m)/2 is even.
To prove the above theorem it is sufficient, in view of Theorem 3 by Vinson [1, p. 42], to prove the following:
Lemma.
divides m.
m = 1 or 2 ors(m)/2 is odd if and only if d\m and 2\f(p) but 4 (///?,/for every odd prime,/?, which
Proof. Let m = 1 or 2 or s(m)/2 be odd. If m = 1 or 2, then the conclusion is clear. So we may assume that m >
2 and s(m)/2 is odd. Suppose %\m. Then by (2), 12 = s(8)\s(m). Therefore s(m)l2 is even, a contradiction. Hence
8f/77.
Let p be any odd prime which divides m. From [1, p. 37] and (2), f(p)\s(p)\s(m). Therefore 4 ]/ f(p). Suppose
2 \f(p). Then by Theorem 1 of [1,p. 39] and (2), we have 4f(p) = s(p)\s(m), a contradiction. Thus 2\f(p).
Conversely, let 8 (/wand 2 \f(p) but 4 \f(p) for every odd prime,/?, which divides m. Let/? be any odd prime which
divides m and let e be any positive integer. From [1, p. 40] we have that f(p) and f(pe) are divisible by the same
power of 2. Therefore 2\f(pe) and %\f(pe). Then since

Binary Sequences Without Isolated Ones: Good A N

Uploaded by

Copyright:

Available Formats

Binary Sequences Without Isolated Ones: Good A N

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Binary Sequences Without Isolated Ones: Good A N

Uploaded by

Copyright:

Available Formats

BINARY SEQUENCES WITHOUT ISOLATED ONES

Hi CHARD AUSTIN and RICHARD GUY

00000, 11000,01100,00110,00011, 11100,01110,00111, 11110,01111, 11011, 11111.

an = 2an,1 - an_2 + an-3

an = ani +an_2 + an-4-

a = 1 -Y2y 0.122561166876, 0 = J2y*-4yl2

an = 2an-i - an2 + an_k-i,

an = ani+ ank + an-k~2 + an-k-3 + '" + an-2k >

BINARY SEQUENCES WITHOUT ISOLATED ONES

- 1), k odd, or mod 2k+1

BINARY SEQUENCES WITHOUT ISOLATED ONES

Fn+r = Fr (mod m) far all integers r if and only if s(m) \n.

m\n implies s(m)\s(n).

It is easily verified that for all integers n

From Theorem 1 of [1, p. 39] we have X\\dXs(m) is even if m > 2.

You might also like