[SciPy-dev] Problem with flapack

[eric]

>
> On Thu, 14 Feb 2002, Pearu Peterson wrote:
>
> > Ok, turns out that the fix is not trivial after all because in January
>
> I was able to figure out a simple fix but I am not sure that the results
> will be correct. The fix is now in scipy CVS. See
>
>
http://scipy.net/cgi-bin/viewcvs.cgi/scipy/linalg/linear_algebra.py.diff?r1=1.13
&r2=1.14
>
> for details. Here is the output of Heiko's test:
>
> >>> from scipy import signal as s
> >>> s.roots([1,2,3,4])
> array([-0.1746854  -1.54686889e+00j, -0.1746854  +1.54686889e+00j,
>             -1.65062919 -3.31070451e-16j])
>
> Is this correct result?

Let see... divide by... carry the two... Yeah.  that looks right. :-)
(Matlab also agrees)

eric