Mathemagix

[Joris van der Hoeven <address@concealed>]

Hi all,

I am happy to announce several improvements in Mathemagix and
its standard C++ template library Mmxlib.

The improvements in Mathemagix itself mainly concern an increase
of robustness. First of all, we now support separate compilation of
the glue code for extern C++ template libraries. This makes it easy
to maintain a Mathemagix interface during the development of
a big library. Secondly, we implemented a first version of a mark
and sweep garbage collector. The GC is still non-incremental and
non-generational, so quite slow, but at least, it is intended to be
far more reliable than Boehm's GC. But if you notice any bugs,
please let me know.

As to the more recent project Mmxlib, the aim is to have a very fast
standard C++ template library for Mathemagix for arithmetic
on dense objects (dense polynomials, dense matrices, power series, etc.).
For my personal research, I also plan to implement the basics of
effective complex analysis into Mmxlib. Finally, there will be some
tools for the computation with symbolic expressions.

Currently, Mmxlib contains:

 * An interface to the GNU multiprecision library gmp and mpfr for basic 
arithmetic.

 * Basic arithmetic for effective real numbers.

 * Several standard template types like lists, hash tables, iterators, etc.

 * Univariate arithmetic for dense polynomials, skew polynomials and series 
(in progress).

 * Dense matrices (in progress).

 * Symbolic expressions.

At this stage, several types of contributions would be very much appreciated:

 * Glue descriptions for various Mathematical C/C++ libraries,
   like Synaps, Linbox, CLN, Ginac, GNU scientific library, etc.

 * A good Emacs-mode for Mathemagix.

 * Any contributions to Mmxlib.

My personal focus for the next improvements should be on implicit
type conversions, generic types and syntactic sugar in Mathemagix,
and starting the implementation of effective complex analysis
inside Mmxlib.

Notice that Mathemagix and Mmxlib are now developed using CVS:

        http://www.mathemagix.org/mmxweb/web/cvs.en.html

Best wishes, Joris