From: Thomas Klimpel (Thomas.Klimpel_at_[hidden])
Date: 2008-05-28 10:57:18
> I do not recall discussions on the functions you want to use. Someone
can confirm this?
I can confirm this. But the discussions can certainly only start after
the request has been made, because lapack contains more than 1000
> Maybe we should change the phrase "on request" in the paper.
> Since this is an open source project and contributions from the
> community are highly appreciated.
I think the phrase communicates at least the problem that there are too
many different functions to provide bindings for all of them. So the
best way to start in case a required binding is missing seems to be to
make a request to the ublas mailing list.
> > Currently I need to calculate eigenvalues and eigenvectors
> > of an upper Hessenberg matrix. This can be accomplished with the
> > following functions
> > HSEQR
> > HSEIN
> > TREVC
A Hessenberg matrix would be a special case of a banded matrix. However,
lapack has no function GBEV (for whatever reasons), so the request seems
reasonable to me. A binding for TREVC should already exist.
I would not object if Jeremy wants to implement the missing bindings
himself, but I certainly want to encourage him to use the bindings
library. So I would be willing to implement the requested bindings.
[mailto:ublas-bounces_at_[hidden]] On Behalf Of Karl Meerbergen
Sent: Mittwoch, 28. Mai 2008 14:57
To: Jeremy Conlin
Cc: ublas mailing list
Subject: Re: [ublas] Adding LAPACK bindings to boost
Such issues are usually discussed in the ublas mailing list (See CC). I
have not followed the latest discussion on the evolution of lapack
bindings. Some functions have been added recently. I do not recall
discussions on the functions you want to use. Someone can confirm this?
Maybe we should change the phrase "on request" in the paper. Since this
is an open source project and contributions from the community are
highly appreciated. I would encourage you to have a look at e.g. gesv()
and getrf() bindings and see whether you are able to write bindings (I
warn you: it is technical). Otherwise I will try and find some time end
Jeremy Conlin wrote:
> Your article (link below) concerning C++ bindings to LAPACK was
> recently brought to my attention. In that article is mentioned that
> additional bindings could be written upon request. May I make a
> request? Currently I need to calculate eigenvalues and eigenvectors
> of an upper Hessenberg matrix. This can be accomplished with the
> following functions
> Those functions take advantage of the upper Hessenberg matrix
> structure. I also need the functions that are used for QR
> factorization and the QR algorithm, but I believe those are already
> implemented in geqrf.hpp
> Thank you,
> Jeremy Conlin