Marketing (Was: Will Free Axiom change our world?)

The evidence to date on commercial macsyma is that it is
not as fast as Maple, generally.  It may or may not be
as fast as Mathematica.  There are ideas that are used
in Maple that were not well known when Macsyma was
designed (in particular, using hashing in some ways).
I've written some programs (rejected by the commercial
macsyma people) to change this balance in the simplifier,
but some of these things can cause subtle problems.


I agree that macsyma/ maxima  could be made rather fast;
I think that all the common lisp implementations allow for
linkages to programs written in other languages (C, Fortran),
with more or less convenience, if you don't trust the CL
itself to do a good job.

There is a point that Macsyma programs themselves (written by
users) can be compiled to lisp or machine language, and so
have a possible advantage over Maple and Mathematica (which does
pattern matching to evaluate commands!)  But the major hit in
speed is often in the execution of commands like "expand" or
"simplify" or "factor".

The big advantage I see in the university environment is the
open code where pieces can be replaced for research purposes
in computer science algorithm improvement;  others may find
that maxima can be used as a front end for very specialized
applications using Lisp, C, etc.  e.g. setting up programs that
do genome matching etc..
RJF


Martin RUBEY wrote:
> On Mon, 21 Apr 2003, James Frye wrote:
> 
> 
>>On Mon, 21 Apr 2003, Martin RUBEY wrote:
>>
>>
>>>...I'm pretty sure that maxima can be made *very* fast.  Simply
>>>monitor your code (this should be possible with metering.lisp), take out
>>>the slow bits and rewrite them in lisp. 
>>
>>Easy enough to say.  Possibly easy to do IF you are one of those people
>>who can understand lisp.  From my own experience, though, that's only a
>>small fraction of the potential user base.
> 
> 
> I think I misstated my opinion. I really should have said: 
> 
> "maxima is very fast and can be made *very* fast".
> 
> For the second part you might not even need lisp: I believe, with gcl you 
> can also use C. But I've no experience with this.
>  
> 
>>I think that may in the long run be the biggest obstacle to maxima's
>>growth: simply finding the people who are both interested and able to work
>>on the project.
> 
> 
> Yes and no, I think. *I* think for those who know and like lisp maxima has 
> a big advantage over other CAS. The others may be equally well served by 
> any other CAS -- disregarding the price. However, at least on university, 
> price is not necessarily the issue: At least in France, MuPad seems to be 
> free for universities, Maple is available. In Austria, MuPad 
> and Mathematica is available. In short: some CAS is always there -- of 
> course, it might well be that you don't like this particular one...
> 
> Martin
> (Please fill out the questionnaire, I'm interested in your opinions!)
> 
> 
> _______________________________________________
> Maxima mailing list
> Maxima@www.math.utexas.edu
> http://www.math.utexas.edu/mailman/listinfo/maxima