From: wgd@berne.ai.mit.edu (Bill Dubuque)
Newsgroups: sci.math.symbolic
Subject: toy computer algebra systems [was: My Program Does REAL Symbolic Calculus]
Date: 11 Sep 1996 10:51:52 -0400
:From: clw@hprnd.rose.hp.com (Carl Wuebker)
:Date: 10 Sep 1996 21:39:51 GMT
:
: ... *Typically*, it's true that S/W written by larger team with several
:years experience in the field is better than one-person-or-small-team-written
:packages ...
I think that perhaps it is not generally appreciated just how enormous
an effort is required to implement a general purpose computer algebra
system. For instance, Macsyma is a product of hundreds of man years of
development (of which I've been part of for almost twenty years).
It is rather naive (to say the least) to believe that an individual
could develop a general purpose computer algebra system that could
compete with the major general purpose computer algebra systems.
While there have been analogous individual accomplishments, e.g.
Knuth's development of TeX, this would require a truly exceptional
individual.
OTOH, it is certainly possible for talented individuals to develop
component modules that would compete with analogous modules in
the commercial systems. Indeed, many commercial component modules
evolved from user contributions. For example, the Macauley system
http://www.math.columbia.edu/~bayer/Macaulay.html, a system for
computation in algebraic geometry and commutative algebra, was
the product of a couple people and surpasses the capabilities
of general purpose systems in its niche of algebraic geometry.
I recommend that anyone interested in developing software for
computer algebra should be sure to become familiar with the
relevant literature. For starters, see the texts of Geddes
et. al. and Zippel. Also see the Journal of Symbolic Computation.
For online resources, start with the following URLs
http://www.can.nl/cain.html
http://symbolicnet.mcs.kent.edu/
Instead of developing toy computer algebra systems, I would
recommend joining a coordinated effort to develop a public
domain computer algebra system. This would be of much greater
value in the long run.
-Bill