[seqfan] Re: Divisible by the digits of a(n-1)

[Benoît Jubin]

It might be interesting to compute the base-3 equivalent of this
sequence, and the asymptotic analysis would be simpler. Indeed, in
base 2, we obtain the identity, and in base 3, the next term of the
sequence is either the smallest integer not yet in the sequence, or
the smallest even integer not yet in the sequence, and the latter case
happens if a "2" appears in the decomposition of the current term
a(n), and there is a probability 1-(2/3)^{log_3 n} ~ 1 - n^{-0.37}
that this happens.

Benoit

On Sat, Feb 15, 2014 at 11:56 AM, Giovanni Resta <g.resta at iit.cnr.it> wrote:
> On 02/15/2014 01:06 AM, Reinhard Zumkeller wrote:
>
>> For A237860 I'm calculating the first 100 terms, a(97) is still running
>> (a(96) = 43).
>
>
> I extended the b-file to 1380 terms (running time about 1 minute, but I had
> to use most of my 32GB of RAM...)
>
> Giovanni
>
>
>
> _______________________________________________
>
> Seqfan Mailing list - http://list.seqfan.eu/