puzzling finding (part.f. and A103446)

[N. J. A. Sloane]

> Ralf, I believe that the terms as shown are correct, but
> that your g.f. is wrong.  Here is the defference
> between what your gf gives and the true values:
> 
> > [seq(s4[i]-t3[i],i=1..40)];
> [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, -1, -31, -498, -5525,
> 
>     -47607, -339762, -2091812, -11427220, -56546963, -257524218, -1093010694]
> 
> So everything seems to be correct (except your gf!)

That explains it but the g.f. WAS correct for all members given
in the OEIS stripped file, so the algorithm is okay, I just had
not enough values!


Thanks for clarification,
ralf