OFFSET
0,7
COMMENTS
a(500) was found by Guo-Gang Gao (see links).
Interestingly, this sequence has many sets of consecutive terms that are increasing powers of 2 minus 1. For example: a(291) to a(307), a(447) to a(467), and a(603) to a(625). It is not clear why this is the case.
The largest value known in this sequence is a(1812) = 2^26-1 = 67108863.
Conjecture: If a(n) = 2^k - 1 for some k > 1, then a(n-1) = 2^(k-1) - 1. Conjecture holds for n <= 1812.
From Hartmut F. W. Hoft, Aug 16 2018: (Start)
The conjecture is true. Let the lengths of the Collatz runs equal q for all numbers 2^n + 1, 2^n + 2, 2^n + 3, 2^n + 4, ..., 2^n + 2^k - 2, 2^n + 2^k - 1. Then dividing the 2^(k-1) - 1 even numbers by two gives rise to the sequence 2^(n-1) + 1, 2^(n-1) + 2, ..., 2^(n-1) + 2^(k-1) - 1 of numbers for which the lengths of the Collatz runs equals q-1. Furthermore, let the length of the Collatz run of 2^n + 2^k be r != q then the length of the Collatz run of 2^(n-1) + 2^(k-1) is r-1 != q-1, i.e., a(n-1) = 2^(k-1) - 1.
Conjecture: Let a(k), ..., a(k+m), m >= 0, be a subsequence of this sequence such that a(k)=a(k+m+1)=1 and a(k+i) > 1, 1 <= i <= m. Then the lengths of the Collatz runs of a(k+i), 0 <= i <= m, increase by 1. In addition, there is an initial segment of increasing numbers a(k), ..., a(k+j), for some 0 <= j <= m, in each such subsequence having the form 2^i - 1, 0 < i <= j. (End)
LINKS
Dmitry Kamenetsky, Table of n, a(n) for n = 0..1812
Dmitry Kamenetsky, C program to compute the sequence
Guo-Gang Gao, On consecutive numbers of the same height in the Collatz problem, Discrete Mathematics, Volume 112, pages 261-267, 1993.
Pureferret, Longest known sequence of identical consecutive Collatz sequence lengths, Mathematics StackExchange, 2013.
Carlos Rivera, Puzzle 847: Consecutive primes with the same Collatz length, The Prime Puzzles and Problems Connection.
Carlos Rivera, Puzzle 851: Puzzle 847 revisited, The Prime Puzzles and Problems Connection.
EXAMPLE
a(6) = 3, because 2^6+1, 2^6+2 and 2^6+3 all take 27 steps to reach 1.
From Hartmut F. W. Hoft, Aug 16 2018: (Start)
Two examples for the conjecture (L(n) denotes the length of the Collatz run):
n a(n) L(n) n a(n) L(n)
64 26 483 20 1 72
------------------ ------------------
65 1 559 21 1 166
66 3 560 22 3 167
67 7 561 23 7 168
68 15 562 24 15 169
69 31 563 ------------------
70 63 564 25 26 170
------------------ 26 26 171
71 26 565 ------------------
72 30 566 27 1 247
73 26 567
74 26 568
75 26 569
76 26 570
77 30 571
78 46 572
79 26 573
80 26 574
81 26 575
82 26 576
------------------
83 1 626
The "power of 2 minus 1" initial section of any such subsequence of a(n) is always increasing. However, there is no apparent ordering in the second section when that is present. (End)
MATHEMATICA
f[n_] := Length[NestWhileList[If[EvenQ@ #, #/2, 3 # + 1] &, n, # != 1 &]] - 1; Table[k = 1; While[f[2^n + k] == f[2^n + k + 1], k++]; k, {n, 120}] (* Michael De Vlieger, Oct 03 2016 *)
PROG
(PARI) nbsteps(n)= s=n; c=0; while(s>1, s=if(s%2, 3*s+1, s/2); c++); c;
a(n) = {my(ns = 2^n+1); my(nbs = nbsteps(ns)); while(nbsteps(ns+1) == nbs, ns++); ns - 2^n; } \\ Michel Marcus, Oct 30 2016
CROSSREFS
KEYWORD
nonn
AUTHOR
Dmitry Kamenetsky, Sep 30 2016
STATUS
approved