Thank you for the solution of first compiling lapack as root - it worked
fine. I had no trouble with this, as I've had to compile code as root for
other software, so it can then be accessible by all users.
My next problem is - are there any Maxima interfaces to the lapack
routines? Suppose, for example, I wanted to calculate the eigenvalues
numerically of a 50x50 matrix. Then:
load(lapack);
M:genmatrix(lambda([i,j],random(1.0)),50,50)$
es:dgeev(M);
But I have to know about the lapack function dgeev to do this. It seems to
me that the lapack routines should be hidden from the average user, so that,
for example
eigenvalues(M),numer;
(or whatever) should call dgeev automatically. In other words, if the user
command calls for a numerical, rather than a symbolic, result, the relevant
lapack routine should be invoked. Or am I missing something here?
cheers,
Alasdair
On 5/7/07, Vadim V. Zhytnikov <vvzhy at mail.ru> wrote:
>
> Robert Dodier writes:
> > On 5/6/07, Vadim V. Zhytnikov <vvzhy at mail.ru> wrote:
> >
> >> I strongly dislike current situation.
> >
> > Well, I'm certainly not happy about it either.
> >
> >>> The only drawback I see is that would increase the size of the
> >>> binary package by some megabytes.
> >> Is it really a lot? These files can be packaged in separate .rpm
> >> and .deb subpackage and or as an extra option in Windows Installer.
> >> I suggest to introduce extra configure option which will control
> >> whether lapack will be compiled or not.
> >
> > The main problem, I think, is that a lot of people are still using
> > dialups, so just to obtain the package is a problem even if there
> > is plenty of room on the disk. A separate rpm or installer helps
> > somewhat by reducing the size of the largest blob that needs to
> > be downloaded.
> >
>
> But optional compilation and separate subpackages
> address this very problem. Even for Windows
> where we have now one monolithic installer
> it is possible to package compiled lapack in
> separate add-on installer.
>
> --
> Vadim V. Zhytnikov
>
> <vvzhy at mail.ru>
> <vvzhy at netorn.ru>
> _______________________________________________
> Maxima mailing list
> Maxima at math.utexas.edu
> http://www.math.utexas.edu/mailman/listinfo/maxima
>