symbolic calculation under scientific programming languages

Richard B. Kreckel kreckel at
Tue Oct 8 15:38:15 CEST 2002


On Mon, 7 Oct 2002, Hervé Tanguy wrote:
> I am deeply interested in using an open-source symbolic computation kernel
> under applications such as Matlab (and to keep open, Scilab and Octave). I
> think this would be an alternative to (very) expensive solutions proposed by
> The Mathworks, to build it upon GiNaC.
> A few questions then:
> * is there a manner of doing this by implementing a kernel usable for the
> three softwares mentionned, and developping specific code only for
> interfacing ?

You would need to write an interface into the library, mapping all the
functionality needed by the frontends to the GiNaC library.  Such an
interface would presumably be a parser, since these frontends want to
communicate with the Kernel through some "language".  For this, you would
have to create a catalogue of requirements and check if everything needed
is available.  This is not going to be the case and filling the holes is
the first task.  This will become quite tedious.

Of course, the idea is not new.  It seems like the MuPAD group is willing
to make the MuPAD language "public" and put their library under an
opensource license.  I have heard words, that some people inside Maple
are suggesting the same for Maple.  But such things have not been decided
yet, as far as I know.

This may turn out to be a slippery road, however: Are those languages good
ones?  And more importantly: Are they stable?  Would it really be possible
to write a replacement kernel for those libaries?  Even if the answer to
all three questions is affirmative, GiNaC is currently not able to fulfill
the needs.

> * are there any legal problems since such a toolbox exists under Matlab
> (built on Maple) ?

I don't know.

(I can only speak for myself here: I don't really want to know.)

> * anyone interested for such developments will express a specific list of
> desiderata (control, maths,...). How to organize the project so as to
> provide anyone with the routines he needs ?

Study the framework, see if it suits your needs, fill in the holes by
submitting code, maybe?

> * What do you think of this ? Does there exist any such project ?

This is not our intention, but if people can substantially contribute to
such a project, such contributions {c|s}ould be incorporated, I assume?

> I am working under Matlab, that's why this proprietary software is
> mentionned here.

That's okay, we are not an "official" GNU website where all mentions of
non-free software must be eliminated from.    ;-)


PS: Please don't format your Email as HTML.
Richard B. Kreckel
<Richard.Kreckel at GiNaC.DE>

More information about the GiNaC-list mailing list