RE: [Axiom-developer] RE: mathaction

[Bill Page]

On Friday, September 17, 2004 11:18 AM Camm Maguire
address@hidden
> 
> Martin Rubey <address@hidden> writes:
> ...
> > Yes, they [Mupad] are *very* strong. One reason for this is
> > that MuPad is free for scientists in France, as far as I know.
> > In fact, if MuPad became free, I think there would be little
> > to no space left for axiom.
> 
> This is our challenge.  Do the mathematicians of the world have
> enough spare brain cycles to move beyond the position of a mere
> consumer choosing among whatever best black box happens to be
> available for the task at hand, to the position of a part-time
> contributor to a system permanently owned by the public, and in
> which the knowledge of *how* something is done is available at
> any time to anyone who desires to read?

Well, one of the goals of MathAction could be to make this
possible with a minimum of concern about programming and
configuring operating systems etc. Certainly in the past
mathematicians have been major contributors to Axiom and Reduce
but the learning curve and the investment in appropriate
infrastructure (including local guru's etc.) is rather
prohibitive. I doubt though, that most people who have used
either of these two systems successfully have really taken
a "black box" view. In general, I don't think computer algebra
is anywhere near mature enough to take such a view - even
among the most developed and commercially funded such as Maple
and Mathematica. In almost all cases I would be quite sure
that sooner or later there was a need to look at the actual
code and often to make improvements in the underlying code
in order to address other specific end results.

The idea to extend this beyond Axiom to include Reduce, Maxima,
Mupad, etc. is a separate but related goal. It seems to me that
because of the very large time and intellectual investments
involved, it has been difficult for many people to take a wider
view and thereby make it possible to better share the critically
limited resources. Maybe having at least a common user interface
(e.g. the collaborative wiki environment such as MathAction)
between the diverse systems will remove one of the barriers to
the pooling of expertise. For example web browser based user
interfaces seem to be quickly growing in popularity among new
Linux users (e.g. Gnome and KDE) and also among large computer
systems administrators (e.g. Webmin for system administration).

> To look at the immense complexity of the Linux kernel, gcc,
> etc., the challenge is not impossible.  But it is critical
> that we prioritize, organize, and gain a community of minds
> *very* familiar with the details of the system.

I hope that becoming a contributor to Axiom (or any other
computer algebra system) does not really require the same
kind of effort as the Linux kernel! If so, then I think the
project is doomed from the start. There just aren't enough
people available to make such a commitment.

>From my point of view, the most promising approach is to
attempt to reduce the *apparent complexity" of these systems
through the use of much improved user interface tools (such
as in the current MathAction experiment ... :) If we assume
that one of the reasons work on these systems has (more or
less) stalled is because they have reached some sort of
human limit in the ability to deal with the complexity, then
it seems reasonable to expect that even a modest reduction
in the apparently complexity would enable many of the
potential contributors to continue the job. Though this is
probably a "too optimistic" scenario.

> I am the worst offender as the only thing in which I'm any
> kind of expert at present is gcl.
>

I don't consider that an "offence"... at least we have you
as an expert! :)
 
> If someone can suggest a good, high priority, somewhat
> manageable task, I'm willing to do my part.
> 

I am not sure what you have in mind but just to get started,
since you did raise the issue of Reduce licensing in a
previous message, may be you would like to try making a
modification to one or more pages on MathAction to reflect
this situation with Reduce? If you have questions about how
to proceed with using the MathAction wiki interface, then
I would be very glad to help - or perhaps Martin also.

Regards,
Bill Page.