%I
%S 0,1,3,6,2,7,13,20,12,21,11,22,10,23,9,24,8,25,43,62,42,63,41,64,40,
%T 65,39,66,38,67,37,68,36,69,35,70,34,71,33,72,32,73,31,74,30,75,29,76,
%U 28,77,27,78,26,79,133,188,132,189,131,190,130,191,129,192,128,193,127,194
%N Lexicographically earliest injective nonnegative sequence a(n) satisfying a(n+1)  a(n) = n for all n.
%C The map n > a(n) is an injective map to the nonnegative integers, i.e., no two terms are identical.
%C Appears not to contain numbers from the following sets (grouped intentionally): {4, 5}, {14, 15, 16, 17, 18, 19}, {44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61}, etc. The numbers of terms in these groups appears to be A008776.  _Paul Raff_, Mar 15 2010
%C The first 23 terms are shared with Recamán's sequence A005132, but from then on they are different.  _Philippe Deléham_, Mar 01 2013, _Omar E. Pol_, Jul 01 2013
%C From _M. F. Hasler_, May 09 2013:
%C It appears that the starting points of the gaps (4, 14, 44, 134, 404, 1214, ...) are given by A181655(2n) = A198643(n1), and thus the ending points (5, 19, 61, ...) by A181655(2n) + A048473(n1).
%C The first differences have signs (grouped intentionally): +++, , +++, ++++ (5 times ""), +++, +...+ (17 times ""), +++, ... where the number of minus signs is again given by A048473 = A0087761. (End)
%C A correspondent, Dennis Reichard, conjectures that (i) a(n) <= 3.5*n for all n and (ii) the sequence covers 2/3 of all natural numbers.  _N. J. A. Sloane_, Jun 30 2018
%H R. Munafo, <a href="http://mrob.com/pub/math/seqa171884.html">Lexicographically earliest injective and unbounded sequence A(n) satisfying A(n+1)A(n)=n for all n</a>
%H R. Munafo, <a href="http://mrob.com/pub/math/mainA171884.txt">mainA171884.c</a>(C source code to generate the sequence)
%F a(n+1) = a(n) + n with  iff n is even but not n = 2 + 2*3^k. (Cf. comment from May 09 2013.)  _M. F. Hasler_, Apr 05 2019
%e We begin with 0, 0+1=1, 1+2=3. 33=0 cannot be the next term because 0 is already in the sequence so we go to 3+3=6. The next could be 64=2 or 6+4=10 but we choose 2 because it is smaller.
%t Contribution from _Paul Raff_, Mar 15 2010: (Start)
%t A171884[{}, _, _] := {};
%t A171884[L_List, max_Integer, True] := If[Length[L] == max, L, With[{n = Length[L]},
%t If[Last[L]  n < 1  MemberQ[L, Last[L]  n],
%t If[MemberQ[L, Last[L] + n],
%t A171884[Drop[L, 1], max, False],
%t A171884[Append[L, Last[L] + n], max, True]],
%t A171884[Append[L, Last[L]  n], max, True]]]]
%t A171884[L_List, max_Integer, False] := With[{n = Length[L]},
%t If[MemberQ[L, Last[L] + n],
%t A171884[Drop[L, 1], max, False],
%t A171884[Append[L, Last[L] + n], max, True]]]
%t A171884[{0}, 200, True]
%t (End)
%o (PARI) A171884_upto(N,a=0,t=2)=vector(N,k, a+=if(!bitand(k,1), k1, t=1, 1k, t=k1)) \\ or:
%o A171884_upto(N,a)=vector(N,k,a+=if(bitand(k,1)&&k\2!=3^valuation(k(k>1),3),1k,k1)) \\ _M. F. Hasler_, Apr 05 2019
%Y Cf. A005132, which allows duplicate values.
%Y Cf. also A118201, in which every value of a(n) and of a(n+1)a(n) occurs exactly once, but does not ensure that the latter is strictly increasing.
%K nonn
%O 1,3
%A _Robert Munafo_, Mar 11 2010
%E Definition edited by _M. F. Hasler_, Apr 01 2019
