Re: silly long list of suggestions

[Joerg Arndt <arndt@jjj.de>]

Thanks for the swift reply.

* Bill Allombert <Bill.Allombert@math.u-bordeaux1.fr> [Apr 15. 2013 18:24]:
> On Mon, Apr 15, 2013 at 05:49:57PM +0200, Joerg Arndt wrote:
> >
> Hello Joerg,
> thanks for your suggestions.
> 
> > Pade approximant for series, see
> >   http://jjj.de/pari/pade.gpi
> > (looks doable)
> 
> Did you try bestappr and bestapprPade ?
> 

I was afraid I'd touch things actually possible already...

? bestapprPade(sin(x),5)
%6 = (911*x^11 - 250734*x^9 + 29604960*x^7 - 1665195840*x^5 + 39178339200*x^3 - 250637587200*x)/(-8981280*x^4 - 2594592000*x^2 - 250637587200)

Good!

> > An option to have polynomials printed "constant first".
> > This is much neater in some cases, e.g., for rational
> > generating functions.
> 
> Well, you can convert your polynomial to a power series for display.

Won't work with fractions of polynomials
(as other things I suggested, there are workarounds;
I only made such suggestions when I thought it would
lead to more expressive Pari code).


> 
> > Option to have Vec(series) _not_ drop initial zeros.
> 
> Try Vec(series,n)

Second arg determines length of vector, not initial index.

I'd like to be able to have
? Vec(x^3/(1-x)+O(x^6))
\\ give
%4 = [0, 0, 0, 1, 1, 1]
\\ and not
%4 = [1, 1, 1]

My usual workaround is
v = Vec(x^3/(1-x)+O(x^6) +'c0);
v[1] -= 'c0;
v
... but that's at least a bit noisy.


> Cheers,
> Bill.

Best,  jj