In One Sequence Or Another (Non-Fibonacci)
Marc LeBrun
mlb at fxpt.com
Fri Mar 26 03:39:32 CET 2004
>=Henry Gould
> I wrote a paper styled "Non-Fibonacci numbers", Fibonacci Quart.,
3(1965), 177-183,
> about the sequence complementary to 1, 2, 3, 5, 8, 13, 21, 34, . . .
A001690, which references your paper.
> and showed how a formula can be gotten.
> ...
Which, alas, is NOT in the OEIS entry. Could you supply it?
>=Leroy Quet
> Do you know what the non-Fibonacci numbers' recursive definition is?
Roughly speaking, a(n+1) = a(n)+1, except when this would produce a
Fibonacci number, in which case it glitches by +2 instead.
> Does their direct formula involve powers of (3 + sqrt(5))/2 somehow?
> (Just a guess. Perhaps the formula does not.)
Well it's basically like n plus the number of Fibonacci numbers less than
n, to provide the requisite glitches.
Thus you need a formula for the density of Fibonaccis (for complement
primes we could use the celebrated prime density).
So I think it will involve log(n) or something like that, somehow twiddled,
roofed and/or summed "in gold".
Alas I don't have time now to take this further; perhaps someone else could
augment A001690 with a formula?
More information about the SeqFan
mailing list