



2, 3, 4, 7, 8, 9, 11, 12, 13, 16, 17, 18, 21, 22, 23, 25, 26, 27, 30, 31, 32, 34, 35, 36, 39, 40, 41, 44, 45, 46, 48, 49, 50, 53, 54, 55, 58, 59, 60, 62, 63, 64, 67, 68, 69, 71, 72, 73, 76, 77, 78, 81, 82, 83, 85, 86, 87, 90, 91, 92, 94, 95, 96, 99, 100, 101
(list;
graph;
refs;
listen;
history;
text;
internal format)



OFFSET

1,1


COMMENTS

Conjecture: 2 < n*r  a(n) < 1 for n >= 1, where r = (7 + sqrt(5))/6. [Corrected by Clark Kimberling, Aug 19 2019]
From Michel Dekking, Aug 22 2019: (Start)
Computation of a formula for a(n): we will show that (a(n)) is a union of three generalized Beatty sequences.
Let T denote the morphism {0>1, 1>000}.
The Fibonacci word xF:=A003849 is a concatenation of the two words v = 100 and w = 10 (ignoring xF(0)). It is known that the order of these words is given by the Fibonacci word itself (cf. A003622). Now note that
T(v) = T(100) = 00011, T(w) = T(10) = 0001.
We see from this that the sequence of first differences of a = A287664 for the first 0 in each triple 000 is given by 5,4,5,5,4,5,4,5,5,4,..., and that in fact these two letters occur as the Fibonacci word on the alphabet {5,4}.
Let {b^(j)} for j = 1,2,3 be the sequence of positions of the jth 0 in the triples 000. From Lemma 8 in the paper by Allouche and Dekking it follows that these sequences are given by
b^(j)(n) = floor(n*phi) + 3*n + r^(j),
where r^(1) = 2, r^(2) = 1, and r^(3) = 0.
Note that a(3*n+j3) = b^(j)(n) for n=1,2,3,... and j=1,2,3. (End)
From Michel Dekking, Aug 22 2019: (Start)
Proof of Kimberling's conjecture. {} denotes fractional part.
Let r = (7+sqrt(5))/6 = phi/3+1. Using the formula above we have
(3*n+j3)*r  a(3*n+j3) =
(n+j/31)*(phi+3)  (n*phi  {n*phi} + 3n + r^(j)) =
(j/31)*phi + j  3  r^(j) + {n*phi}.
Since ({n*phi}) is equidistributed on (0,1), this implies that a(3*n+j3) takes values in the interval ((2/3)*phi, (2/3)*phi+1) for j=1, in the interval ((1/3)*phi, (1/3)*phi+1) for j=2, and in the interval (0,1) for j=3. Combining these, we see that a(n) takes values in ((2/3)*phi, 1), and that this is best possible. Since (2/3)*phi = 1.078689..., this improves on the lower bound conjectured by Kimberling.
(End)


LINKS

Clark Kimberling, Table of n, a(n) for n = 1..10000
J.P. Allouche, F. M. Dekking, Generalized Beatty sequences and complementary triples, arXiv:1809.03424 [math.NT], 2018.


FORMULA

a(3*n+j3) = b^(j)(n) for j=1,2,3, with b^(1)(n) = floor(n*phi) + 3*n  2, b^(2)(n) = floor(n*phi) + 3*n  1, b^(3)(n) = floor(n*phi) + 3*n.
 Michel Dekking, Aug 22 2019


MATHEMATICA

s = Nest[Flatten[# /. {0 > {0, 1}, 1 > {0}}] &, {0}, 10] (* A003849 *)
w = StringJoin[Map[ToString, s]]
w1 = StringReplace[w, {"0" > "1", "1" > "000"}]
st = ToCharacterCode[w1]  48 (* A287663 *)
Flatten[Position[st, 0]] (* A287664 *)
Flatten[Position[st, 1]] (* A287665 *)


CROSSREFS

Cf. A287663, A287665.
Sequence in context: A070942 A073798 A124375 * A037080 A167055 A022162
Adjacent sequences: A287661 A287662 A287663 * A287665 A287666 A287667


KEYWORD

nonn,easy


AUTHOR

Clark Kimberling, Jun 02 2017


STATUS

approved



