[seqfan] Re: could GUESSS, which superseeker uses, be improved ?

[Olivier Gerard]

Dear Alexander,

This would not be an improvement but a potential limitation.

You are free to discard fractional results yourself
if you have hints that the sequence is strictly integral.

In the case you have tested, it just means that a gf is
very improbable unless it is much more complicated
that what GUESSS is designed and parametrized to look for.

For other users and other sequences, a fractional continuation
could be a valuable insight about the nature of the sequence.

Regards,

Olivier


On Mon, Aug 29, 2011 at 13:11, Alexander P-sky <apovolot at gmail.com> wrote:
> Could "GUESSS", HARM DERKSEN'S PROGRAM FOR GUESSING A GENERATING
> FUNCTION FOR A SEQUENCE, which superseeker uses, be improved to
> exclude cases with non-integer results as shown below ?
> Any taker ?
>
> Thanks,
> ARP
>
> ---------- Forwarded message ----------
> From: superseq-reply at oeis.org
>
> Report on [ 3,5,7,8,10,11,12,12,14,15,16,16,18,19,20,20,21,21]:
>
> TRY "GUESSS", HARM DERKSEN'S PROGRAM FOR GUESSING A GENERATING
> FUNCTION FOR A SEQUENCE.
>
> Guesss suggests that the generating function  F(x)
> may satisfy the following algebraic or differential equation:
>
> -71/64*x^9-69/32*x^8-441/64*x^7-145/16*x^6-525/64*x^5-467/64*x^4-113/64*x^3+2*x
> ^2+2*x+3+(x^9-x^8-71/64*x^7+1/16*x^6-27/16*x^5+25/16*x^4+59/64*x^3+x-1)*F(x) =
> 0
>
> If this is correct the next 6 numbers in the sequence are:
>
> [21, 1221/64, 1003/64, 335/32, 14567/4096, -9839/1024]
>
> _______________________________________________
>
> Seqfan Mailing list - http://list.seqfan.eu/
>