[SciPy-dev] Problem with flapack

[eric]

> thank you for fixing the problem. Here the result from a short test:
>
> >>> from scipy import *
> >>> from scipy import signal as s
> >>> s.poly(s.roots([1.,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20]))
> array([  1.+0.00000000e+000j,   2.-1.89030962e-015j,   3.+3.97294109e-015j,
>               4.+2.16188134e-015j,   5.-5.21090017e-014j,
>               6.-1.44872571e-013j,   7.-2.18532601e-013j,
>               8.-2.36476441e-013j,   9.-2.38065611e-013j,
>              10.-4.36395620e-013j,  11.-9.58807728e-013j,
>              12.-1.45392938e-012j,  13.-1.52765658e-012j,
>              14.-1.21333755e-012j,  15.-6.78134691e-013j,
>              16.-2.67591593e-013j,  17.-7.35574790e-014j,
>              18.-8.47677514e-014j,  19.-1.43588852e-013j,
>              20.-9.54262406e-014j])
> >>>
>
>
> It works fine. It should be possible to polish the values, in order to get
> rid of the small imaginary error, in the function poly.

Yeah, I noticed that too.  I thought we already had done this???  Didn't Travis
O. put this in at one point.  Maybe it was on a different set of functions.
Anyway, this should be added...

eric