Would it be better to just check for floating-point powers, and refuse to do
this?
Insisting that the exponents in polynomials be less than 2^30 or so, and
have the
system run 50% faster, might not be a bad tradeoff. However, I just made up
that 50%,
and the ratio might be much less, especially on lisps that are already slow
in doing
arithmetic optimization.
This would only be for polynomials undergoing "ratsimp" or "radcan", which
might include
stuff going through "solve".
RJF
> -----Original Message-----
> From: maxima-bounces at math.utexas.edu
> [mailto:maxima-bounces at math.utexas.edu] On Behalf Of Stavros Macrakis
> Sent: Thursday, November 29, 2007 4:19 PM
> To: Andreas Eder
> Cc: maxima at math.utexas.edu; Robert Dodier
> Subject: Re: [Maxima] radcan error
>
> On Nov 29, 2007 6:02 PM, Andreas Eder <aeder at arcor.de> wrote:
> >
> > Well, radcan(x^0.000000001/x^0.00000001); worked for me in sbcl
> > and clisp on an 64-bit platform and in gcl on a 32 bit platform.
> > But in cmucl on 32-bit it errors out with:
>
> A simpler test case, which also fails on 32-bit GCL, is
> radcan(x^1.0e-20);
>
> -s
> _______________________________________________
> Maxima mailing list
> Maxima at math.utexas.edu
> http://www.math.utexas.edu/mailman/listinfo/maxima
>