[seqfan] Re: Two neighbors sum -- and odd ranks in S

[M. F. Hasler]

> No need to apologize!  It's amazing to me that Haskell programs run at all
> - magic!

I agree with both statements. :-)

Reinhard, I had indeed a similar but in a first version of my program,
which I cured by making sure that as soon as some a(n) is fixed (in
particular also a(2k)), then I "reserve" the value a(n-1)+a(n) as well
as a(n)+a(n+1) in case the latter already is defined. I think that not
doing the second part led to duplicate values.

Thanks to Alois for the big b-file which yields a graph that is almost
a straight line.
I think one can establish a very small bound for d(n) = a(n)-n:
it seems that values d(n) = k > 0 occur first around  n ~ 1000*2^(k-4)
(for k>3),
and the value -k then occurs first at index n+2k-1 (for all k>0 !?).

(I added this as comment to http://oeis.org/A249129 .)

--Maximilian