From: Richard Fateman
Subject: Re: Axiom or Maxima?
Date:
Message-ID: <h2dntb$3g8$1@aioe.org>
Francogrex wrote:
> This may be a bit offtopic
yes.
since it's not related to a technical CL
> question: There are those 2 computer algebra systems: OpenAxiom and
> Maxima; both are based on Lisp.
Yes. There is also Reduce, Jacal, and probably other systems.
I wanted to know whether you have a
> preference/recommendation for one or the other.
Yes.
Which one of the two
> do you think is a more powerful/mature CAS (based on the scope and
> strength of solving a wide application of mathematical problems)?
Maxima. But you have not defined your criteria sufficiently.
> Which is at the same level as Wolfram's Mathematica?
Again, you have not defined your criterion. Mathematica is clearly far
far above the level of Maxima and Axiom in terms of price. Oh, maybe you
meant something else? Why don't you make a study of this and tell us
what you think.
And last but not
> least which one of the two is based on more ANSI compliant Common
> Lisp?
Why do you ask? How do you grade (partial?) compliance? Many people use
CAS through front ends that are not written in Lisp at all, but in C,
Tcl, emacs-lisp, TeX, etc.
If you do not care about such things, but are trying to slurp up some
computer algebra system in some larger (lisp-based) application, why
don't you just say what you are trying to do instead of asking people to
read your mind?
On Jun 30, 9:09 pm, Richard Fateman <·······@cs.berkeley.edu> wrote:
> Francogrex wrote:
> If you do not care about such things, but are trying to slurp up some
> computer algebra system in some larger (lisp-based) application, why
> don't you just say what you are trying to do instead of asking people to
> read your mind?
I do care, but I may also introduce some CAS in a lisp project (for
statistics), yes, though haven't started on that yet.
Madhu wrote:
> * Richard Fateman <············@aioe.org> :
> Wrote on Tue, 30 Jun 2009 12:09:11 -0700:
>
> | Yes. There is also Reduce, Jacal, and probably other systems.
>
> Do you have any explanations or insights why the approach taken by Weyl
> (representing algebra with CLOS types and generic functions) was not
> taken up by any other CAS system implemented in CL ?
Weyl was a system design by Rich Zippel, who I mentioned earlier.
http://historical.ncstrl.org/tr/temp/ocr/cs-tr.cs.cornell.edu/TR90-1077
For a CAS to "take up" the approach from Weyl would require a
substantial amount of work, for which they would get the following:
(a) A system that was no faster or efficient than before. And maybe slower.
(b) A system with flexibility that existing users would be unaware of
and probably poorly equipped to exploit.
Looking toward new more ambitious applications, Weyl might shine.
As I recall, Zippel sought funding from DARPA for what appeared (to
me, at least) to be a very interesting proposal. Not funded.
The proposal reappeared later but with an implementation in a different
language (maybe C++?), which perhaps gives a hint as to what DARPA said
to him when rejecting the first Weyl proposal. They rejected the second
one, also. There may, of course, be more complex reasons for this.
Perhaps someone with more direct knowledge of the situation can correct
me. (e.g. Rich Z/?)
RJF
On 1 Jul., 16:10, Richard Fateman <·······@cs.berkeley.edu> wrote:
> Madhu wrote:
> > * Richard Fateman <············@aioe.org> :
> > Wrote on Tue, 30 Jun 2009 12:09:11 -0700:
>
> > | Yes. There is also Reduce, Jacal, and probably other systems.
>
> > Do you have any explanations or insights why the approach taken by Weyl
> > (representing algebra with CLOS types and generic functions) was not
> > taken up by any other CAS system implemented in CL ?
>
> Weyl was a system design by Rich Zippel, who I mentioned earlier.http://historical.ncstrl.org/tr/temp/ocr/cs-tr.cs.cornell.edu/TR90-1077
>
> For a CAS to "take up" the approach from Weyl would require a
> substantial amount of work, for which they would get the following:
> (a) A system that was no faster or efficient than before. And maybe slower.
> (b) A system with flexibility that existing users would be unaware of
> and probably poorly equipped to exploit.
>
> Looking toward new more ambitious applications, Weyl might shine.
>
> As I recall, Zippel sought funding from DARPA for what appeared (to
> me, at least) to be a very interesting proposal. Not funded.
>
Wouldn't it be great if the USA could fund computer science and
mathematics from non-defense-related research agencies? This would
also help to ensure that the developed technologies survive in the
marketplace a bit longer.
Kind of tragic that the USA wastes so much engineering brain cells on
military research. It helped create the AI bubble and also helped to
ensure that the AI technology was too expensive, over-engineered and
without civil use.
For Lisp and AI DARPA was (and still is) the kiss of death.
Add to that the questionable size of the military-industrial complex.
Isn't there any funding for CAS research outside of DARPA?
> The proposal reappeared later but with an implementation in a different
> language (maybe C++?), which perhaps gives a hint as to what DARPA said
> to him when rejecting the first Weyl proposal. They rejected the second
> one, also. There may, of course, be more complex reasons for this.
>
> Perhaps someone with more direct knowledge of the situation can correct
> me. (e.g. Rich Z/?)
>
> RJF
Rainer Joswig wrote:
...
>
> Wouldn't it be great if the USA could fund computer science and
> mathematics from non-defense-related research agencies? This would
> also help to ensure that the developed technologies survive in the
> marketplace a bit longer.
There is funding from the National Science Foundation. Unfortunately the
amount of money and the percentage of proposals awarded grants is rather
low (10%?)
Consider also the fact that the grant must (generally) pay for the
tuition and living expenses of the graduate students, and often the
summer salary of the faculty. Thus each student costs, after all is
covered, maybe $60,000 per year. Additional costs for faculty research
time, computer maintenance, travel, staff, etc. run the cost to
$100,000. That is a minimum. Figure $250,000/year for a few years.
Compare to typical European university (or Canada!) where the grant
pays for "extra expenses" but the salaries and tuition are part of the
educational grant. Running a project might cost the equivalent of
$30,000(US)/year.
>
> Kind of tragic that the USA wastes so much engineering brain cells on
> military research. It helped create the AI bubble and also helped to
> ensure that the AI technology was too expensive, over-engineered and
> without civil use.
I think that AI is, in fact, used in non-military applications. Probably
far more than in military. And AI research spawned huge numbers of very
practical things, (relevant to comp.lang.lisp -- compiler/language
research, and Common Lisp ...) and of course military funding lead to
things like the internet, supercomputers, integrated circuits.. So it is
not ALL wasted :) It is, I think, easier to sell military ideas to
voters. E.g. Dare we fall behind in missile defense? .. Compare this to
bemoaning the international computer-algebra gap.
>
> For Lisp and AI DARPA was (and still is) the kiss of death.
DARPA was a major contributor to the early growth of AI. DARPA just
moved on, as it does from time to time. If some of the companies could
live only from defense money, then whose fault was that? (Thinking
Machines and Symbolics come to mind...)
>
> Add to that the questionable size of the military-industrial complex.
Well, if it were smaller, I doubt that would increase DARPA funding for
AI!
>
> Isn't there any funding for CAS research outside of DARPA?
I think that ALL the funding for CAS research is outside of DARPA.
Some of it is commercial (e.g. Maplesoft, WRI, Matlab). Some of it is
other government agencies (in US, that would be National Science
Foundation, perhaps Dept. of Energy or National Security Agency ...)
I got some money years ago from a private foundation (SDF) and CAS
researchers report occasional grants from companies like IBM, HP, DEC,
Sun. And in other countries, esp. Canada, there seems to be more
sympathy for CAS research.
Rainer Joswig <······@lisp.de> writes:
>
> Wouldn't it be great if the USA could fund computer science and
> mathematics from non-defense-related research agencies?
The problem is that most things not military are outside the bailiwick
of the United States. DARPA is constitutional since Article I, Section
8 gives the Congress the power to support armies and maintain a navy.
If you read Article I, the federal government really doesn't have an
awful lot of power.
Now, the individual states are of course free to fund that research
however they wish to.
--
There is nothing more irksome than self-appointed do-gooders, who're so
convinced of the intellectual and moral inferiority of the rest of
humankind that they feel compelled to protect the miserable masses from
their own depravity and stupidity. --Stefaan A. Eeckels
Robert Uhl <·········@NOSPAMgmail.com> writes:
> Rainer Joswig <······@lisp.de> writes:
> >
> > Wouldn't it be great if the USA could fund computer science and
> > mathematics from non-defense-related research agencies?
>
> The problem is that most things not military are outside the bailiwick
> of the United States. DARPA is constitutional since Article I, Section
> 8 gives the Congress the power to support armies and maintain a navy.
> If you read Article I, the federal government really doesn't have an
> awful lot of power.
>
> Now, the individual states are of course free to fund that research
> however they wish to.
I don't buy this argument.
Certainly the NSF and NIH fund a large amount of scientific and medical
related research without running afoul of the U.S. Constitution.
Besides, the Interstate Commerce Clause has for the last century or so
been a loophole large enough, if you'll pardon the pun, to drive a truck
through.
This is more of a philosophical or political issue, with a desire on the
part of some elements in US politics not to engage in activities that
might be construed as "industrial policy", and that often leads to
some programs being run more out of the Defense Department instead of
other cabinet agencies.
NSF = National Science Foundation
NIH = National Institutes of Health
--
Thomas A. Russ, USC/Information Sciences Institute