OFFSET

1,2

COMMENTS

The corresponding values of k are given in A352919.

This is a subset of A352336.

It is not necessary for a term of this sequence to be of the form 2^k - 1: there may be a zero close to the end of the binary expansion.

It appears that n/A109812(n) is unbounded. The reasoning behind this is as follows.

Consider terms A109812(k) that are the form 2^i - 1 (see the Examples section).

For such k, we necessarily have

a(k+1) = p(i)*2^i and a(k-1) = m(i)*2^i,

with integers p(i) and m(i). Let r(i) = max(p(i), m(i)).

Taking A109812(0) = 0, we have the following values:

i : 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

p(i): 1 2 3 4 5 7 7 9 9 11 12 13 13 15 15 17 17 19 [A352921]

m(i): 0 1 4 3 6 6 8 8 10 10 11 14 14 16 18 18 18 20 [A352922]

r(i): 1 2 4 4 6 7 8 9 10 11 12 14 14 16 18 18 18 20 [A352923]

for i < 19. Furthermore, from the graphs in the entry A109812 it appears that r(19) = 21, r(20) = 22, r(21) = 22, r(22) = 24. The corresponding four values of k/aA109812(k) are, approximately, 6.42199, 6.80074, 6.88852, 7.39979.

This suggests the following conjecture:

Conjecture: r(k) > k for all k > 4.

Conjecture: k/A109812(k) is unbounded.

EXAMPLE

Let c(k) denote A109812(k). The first 14 record high-points of k/c(k) are as follows:

[k/c(k), k, c(k), "binary(c(n))"]

[1.000000000 1 1 "1"]

[1.333333333 4 3 "11"]

[1.500000000 9 6 "110"]

[2.285714286 16 7 "111"]

[2.451612903 76 31 "11111"]

[2.571428571 162 63 "111111"]

[3.291338583 418 127 "1111111"]

[3.702544031 1892 511 "111111111"]

[4.665037870 19094 4093 "111111111101"]

[4.713727406 19298 4094 "111111111110"]

[4.898412698 20059 4095 "111111111111"]

[5.167124458 84653 16383 "11111111111111"]

[5.327494125 174566 32767 "111111111111111"]

[6.439611205 1688099 262143 "111111111111111111"]

The values of k and c(k) form A352919 and the present sequence.

CROSSREFS

KEYWORD

nonn,more

AUTHOR

David Broadhurst, Aug 17 2022 (entry created by N. J. A. Sloane, Apr 23 2022)

STATUS

approved