Yes,
I undestood that expand is bad.
What I need is something asymptotically faster
than O(n^2) because I must do some benchmarks
for which I need a better than O(n^2)
multiplication and division algorithm.
Is "*" for CRE polynomials using (standard univariate) Karatsuba?
If not, is (standard univariate) Karatsuba implemented
in Maxima?
Fabrizio
On Wed, 8 Feb 2006, Stavros Macrakis wrote:
> > In all my tests "fasttimes" is much faster than doing
> > the product of two polynomials by "expand(...*...)"
> > even if the polynomials are univariate.
>
> Expand uses general representation, which is very slow for polynomial
> calculations. CRE multiplication is much faster than expand, so you
> should write rat(...)*rat(...) or rat(...*...). Even better, just
> always keep the polynomials in CRE form. Divide works on CREs.
>
> -s
>