Dev-C++ and eigenvectors/values

116 views
Skip to first unread message

jski

unread,
Dec 4, 2010, 12:58:56 AM12/4/10
to matrixprogramming
I've recently begun working with a group of MEs (mechanical engineers)
who've got some rather large matrices (5K x 5K) that they need to
extract the eigenvector form. They work with Dev-C++ and want to stay
there (i.e., Cygwin ain't an option). The only library I know of that
can work with matrices of this size is LAPACK. I've downloaded
(http://sourceforge.net/projects/gnuwin32/files/gsl/) the GLS (which
includes both CBLAS & BLAS) but there's eigenvector/values routines in
there. So as I see it, I've got 2 options:

1) Download the pre-built LAPACK.lib, use reimp.exe to convert it to
LAPACK.a, and try to use this?

2) Try to build *.a file from scratch using MinGW (not sure how)?

Is anyone using Devc-C++ with LINPACK (or CLINPACK)? How did you
accomplish this?

---jski

Evgenii Rudnyi

unread,
Dec 4, 2010, 4:31:52 AM12/4/10
to matrixpr...@googlegroups.com
on 04.12.2010 06:58 jski said the following:

> I've recently begun working with a group of MEs (mechanical
> engineers) who've got some rather large matrices (5K x 5K) that they
> need to extract the eigenvector form. They work with Dev-C++ and
> want to stay there (i.e., Cygwin ain't an option). The only library
> I know of that can work with matrices of this size is LAPACK. I've
> downloaded (http://sourceforge.net/projects/gnuwin32/files/gsl/) the
> GLS (which includes both CBLAS& BLAS) but there's eigenvector/values

> routines in there. So as I see it, I've got 2 options:
>
> 1) Download the pre-built LAPACK.lib, use reimp.exe to convert it to
> LAPACK.a, and try to use this?
>
> 2) Try to build *.a file from scratch using MinGW (not sure how)?
>
> Is anyone using Devc-C++ with LINPACK (or CLINPACK)? How did you
> accomplish this?

You need a Fortran compiler to compile LAPACK. Provided that you have
it, it is relative easy, basically just run make especially if you use
the reference BLAS supplied with LAPACK.

The main problem here is that the reference BLAS is slow and it makes
sense to use of the optimized BLAS. Here you may already have some more
problems.

Reply all
Reply to author
Forward
0 new messages