New to gsl (Re: multidimensional optimization)

[Theodore Papadopoulo <Theodore dot Papadopoulo at sophia dot inria dot fr>]

	Hi,

I just discovered gsl and I think it is a great idea...

For now, I have just some random remarks:

1) The directions to get the sources through cvs are wrong...
   It should be anoncvs@sourceware.cygnus.com instead of
   anoncvs@anoncvs.cygnus.com unless I have been hit by a transient
   error.

2) I have seen no conclusion about the C/C++ discussion. If this is
   still an open issue, then I'll definitely vote for C++ (mainly
   because the use of templates can simplify a lot the maintenance
   and the power of numerical libraries). If there is still interest
   in this issue, I'll expand on this later. I understand the C
   programmers point of view, but I guess this is one field where
   it should be better to develop in C++ and provide C wrappers !!!

3) As for multidimensionnal optimization, I have rewritten in C++
   a package called minpack that does Levenberg-Marquardt 
   optimization. If you go along with C++, I might (up to some 
   administrative checks) provide it to GSL if there is interest for
   that. At the very least, it can be a basis...
   However, it's not yet perfect I must say (in terms of efficiency).
   For now, it uses vectors and matrices based either on a package
   called blitz or on another one called tnt (heavily hacked to
   provide some submatrices and subvector functionnalities), but it
   can be adapted to any other reasonnable matrix/vector scheme.

I just downloaded the sources and I'll have a look at those.

	Thank's for your attention,

	Theo.

 --------------------------------------------------------------------
 Theodore Papadopoulo
 Email: Theodore.Papadopoulo@sophia.inria.fr Tel: (33) 04 92 38 76 01
 --------------------------------------------------------------------