[SciPy-user] RREF - upper triangular

Arnd Baecker arnd.baecker at web.de
Thu Feb 23 04:24:38 CST 2006


Hi Nils,

On Thu, 23 Feb 2006, Nils Wagner wrote:

> arnd.baecker at web.de wrote:
> >On Wed, 22 Feb 2006, Alan G Isaac wrote:
> >
> >>On Wed, 22 Feb 2006, Nils Wagner apparently wrote:
> >>
> >>>here is a code. However it doesn't make use of LAPACK. So
> >>>it might be to slow.
> >>>
> >>This exchange suggests to me that Robert thinks that
> >>wrapping the LAPACK code is approximately trivial but that
> >>Nils does not see it that way.  Apologies for any
> >>misunderstanding.
> >>
> >But in this particular case previous exposure to wrapping lapack
> >functions is granted:
> >http://aspn.activestate.com/ASPN/Mail/Message/scipy-user/2016895
> >(for dgehrd)

[... snipped my remarks on how to use f2py ...]

> If it is so trivial to add the rank revealing QR decomposition to scipy,
> I would be glad if the several steps on building a wrapper for DGEQP3
> could be documented on the Wiki as an exercise.
> The best way to learn something is to have a well-elaborated example. :-)

Have you *studied* the material I wrote in the snipped part,
*including* the texts I referred to?

Also, maybe you remember, that the wrapper *you* posted
at http://aspn.activestate.com/ASPN/Mail/Message/scipy-user/201689
and claimed to be *yours* was in fact written by *me* for *you*.
We had an off-list exchange with over 20 mails on this
in which I tried to guide you step-by-step
through the Hessenberg wrapping. In course
of that I also send you a detailed step-by-step example
on how to wrap band matrices.

So the exercise you propose (including the addition to the wiki)
is really left to you.
I am sure that this list will help you with any concrete questions.

I better stop now, before I write things which I might regret later ...

Arnd



More information about the SciPy-user mailing list