any Python equivalent of Math::Polynomial::Solve?

[Just]

In article <1109485184.587035.176540 at l41g2000cwc.googlegroups.com>,
 "Carl Banks" <invalidemail at aerojockey.com> wrote:

> Just wrote:
> > While googling for a non-linear equation solver, I found
> > Math::Polynomial::Solve in CPAN. It seems a great little module,
> except
> > it's not Python... I'm especially looking for its poly_root()
> > functionality (which solves arbitrary polynomials). Does anyone know
> of
> > a Python module/package that implements that?
> 
> If you don't have a great need for speed, you can accomplish this
> easily with the linear algebra module of Numeric/numarray.   Suppose
> your quintic polynomial's in the form
> 
>    a + b*x + c*x**2 + d*x**3 + e*x**4 + x**5
> 
> The roots of it are equal to the eigenvalues of the companion matrix:
> 
>   0   1   0   0   0
>   0   0   1   0   0
>   0   0   0   1   0
>   0   0   0   0   1
>  -a  -b  -c  -d  -e
> 
> It should be pretty easy to set up a Numeric matrix and call
> LinearAlgebra.eigenvalues.  For example, here is a simple quintic
> solver:
> 
> . from Numeric import *
> . from LinearAlgebra import *
> .
> . def quinticroots(p):
> .     cm = zeros((5,5),Float32)
> .     cm[0,1] = cm[1,2] = cm[2,3] = cm[3,4] = 1.0
> .     cm[4,0] = -p[0]
> .     cm[4,1] = -p[1]
> .     cm[4,2] = -p[2]
> .     cm[4,3] = -p[3]
> .     cm[4,4] = -p[4]
> .     return eigenvalues(cm)
> 
> 
> now-you-can-find-all-five-Lagrange-points-ly yr's,

Wow, THANKS. This was the answer I was secretly hoping for... "Great 
need for speed", no, not really, but this Numeric-based version is about 
9 times faster than what I translated from Perl code yesterday, so from 
where I'm standing your version is blazingly fast...

Thanks again,

Just