Re: [Getfem-users] collaboration between GMM and KDE/Eigen

[Yves Renard]

Le Dimanche 21 Janvier 2007 23:04, Benoît Jacob a écrit :
> (re-sending because I forgot to CC Ben the first time. Please ignore
> previous e-mail. Sorry about that.)
>
> Hello,
>
> This mail is to propose a collaboration between the GMM project and parts
> of the KDE project.
>
> Several subprojects of KDE (kde.org) have identified a common need for
> linear algebra functionality. After having considered the existing linear
> algebra libraries, we found that we needed to start a new one, which we
> called Eigen (eigen.tuxfamily.org). Like GMM, Eigen is a C++ template
> library.
>
> More recently, a new need has appeared in Krita (koffice.org/krita), that
> Eigen can't currently fulfill: sparse matrices and vectors.
>
> Since we see no point in reinventing the wheel, we would like to reuse
> existing code for that. After examining the existing libraries, we found
> that GMM was the most interesting one from our point of view.
>
> However, using GMM as an external dependency is not optimal for
> us, for several reasons:
> - additional dependencies are frowned upon by distribution packagers
> - we are only interested in a small subset of GMM, and GMM doesn't cover
> all of our needs
> - we wouldn't like KDE to depend on two different incompatible linear
> algebra libraries, especially as there is much feature overlap between
> them.
>
> So the best solution for us would be, if you allowed us to copy GMM code
> directly into Eigen. Eigen is licensed under the GPL plus an exception
> allowing any software to use it (much like the GNU C++ STL license, except
> that our exception is more liberal). This is very similar in spirit to the
> LGPL used by GMM. Of course, if we copy GMM code into Eigen, we will
> preserve the GMM authors's copyright lines. But the license will become the
> GPL+exception license used by Eigen.
>
> So do you allow us to copy GMM code into Eigen? If yes, we'd probably have
> a few questions to ask about the GMM source code.
>
> Conversely, we are planning to implement sparse algorithms that aren't yet
> available in GMM, namely sparse least squares:
>
> http://www.stanford.edu/group/SOL/software/lsqr.html
>
> We'd then be happy to back-port them to GMM if you're interested. This way,
> our collaboration would be reciprocal :)

There is of course no problem to import parts of Gmm++ code into a free 
project like Eigen and distribute them with the GPL+exception license used by 
this project.

Questions are welcome. Some parts of Gmm++ are not completely limpid, 
especially on the definition of structures referencing sparse object ... !

Sincerly,

Yves.

-------------------------------------------------------------------------
  Yves Renard (address@hidden)       tel : (33) 04.72.43.80.11
  Pole de Mathematiques, INSA de Lyon          fax : (33) 04.72.43.85.29
  Institut Camille Jordan - CNRS UMR 5208
  20, rue Albert Einstein
  69621 Villeurbanne Cedex, FRANCE
  http://math.univ-lyon1.fr/~renard
-------------------------------------------------------------------------