[seqfan] Re: Binary Complement Sequences

[Tim Peters]

[me]
> The first implies that, modulo 2**n, the -3*i-1 map is injective, and
> the second that it's also surjective. So it's a permutation:
> regardless of n, the last n bits fall into a cycle of length 2**n.

Sorry about that! That the map defines a permutation does not imply
it's a single cycle.

In the background, a quick program confirmed that it is for n though
25, but offhand I'm missing a simple proof.