%I #12 Jan 04 2018 03:56:18
%S 79,83,92,101,176,258,425,445,518,679,733,758,839,932,1272,1311,1382,
%T 1432,1453,1454,1495,1899,1904,1991,2207,2317,2444,2450,2570,2609,
%U 2733,2736,2773,2783,2961,3038,3076,3166,3291,3370,3384,3416,3456,3473,3539,3545
%N Numbers n such that the sequence defined by {b(1)=1, b(2)=n, b(k) = squarefree part of b(k-1)+b(k-2) for k>2} does not eventually merge with the period-6 sequence A272636.
%C The sequences b(n) for the initial terms 79, 83, 92, 101, 176, 258, 425, 445 all merge with the period-43 sequence A272637. In fact for n=83 b(n) IS A272637.
%C From _Lars Blomberg_, Jan 02 2018: (Start)
%C Among the first 10000 terms, two additional cycles appear:
%C A) with period 22: [15, 2162, 2177, 4339, 181, 1130, 1311, 2441, 938, 3379, 4317, 481, 4798, 5279, 10077, 3839, 71, 3910, 3981, 7891, 742, 8633].
%C B) with period 26: [23, 1801, 114, 1915, 2029, 986, 335, 1321, 46, 1367, 157, 381, 538, 919, 1457, 66, 1523, 1589, 778, 263, 1041, 326, 1367, 1693, 85, 1778].
%C The first occurrence of cycle A is n=28243, for cycle B n=21245.
%C Cycle A occurs for 17 terms, B for 24 terms and A272637 for the remaining 9959 terms. (End)
%H Lars Blomberg, <a href="/A272639/b272639.txt">Table of n, a(n) for n = 1..10000</a>
%Y Cf. A007913, A272636, A272637, A272638.
%K nonn
%O 1,1
%A _N. J. A. Sloane_, May 05 2016
%E Terms a(9) and beyond from _Lars Blomberg_, Jan 02 2018