Name: Rolf Mertig Date: 12/09/04-01:56:12 AM Z


Hi,
maybe this is not so well known, but I wrote the first FeynCalc version
completely in Macsyma in
1988-1989 (my Diploma thesis in theoretical physics) and switched to
Mathematica in 1990.

I still do have the original code printouts and even a MicroVAX tape
which should have the code on it,
unfortunately I do not have a MicroVAX, maybe someone in Germany (close
to Berlin) has one).
However, that code is pretty old and far from sufficient for actual
research calculations.
The last time I checked the open-source Maxima for speed it was at least
an order of magnitude
slower for plain algebraic simplifications. Bearing in mind that even
Mathematca can become way too
slow for research projects (see the enourmous calculations of Jos
Vermaseren and friends which they
have done over the last decades, using a special purpose system: FORM),
I really doubt that it will make
much sense to reprogram much of FeynCalc into Maxima.
Of course, hardware gets faster, and maybe some students would like to
do something nice in open source,
but it would make much more sense to write a wrapper from Maxima to FORM
(like Thomas Hahn did once
for Mathematica and FORM, and I even wrote simple FeynCalc2FORM and
FORM2FeynCalc functions some day).

It is, to my mind, basically useless, if non-research-physicists write
large application programs like FeynCalc, FeynArts,
FORM and alike (all of those have been written by physicists …).
Of course it is very useful if new multivariate factorization algorithms
are implemented into existing CA-systems, and if mathematicicans
develop new algorirthms with fancy mathematica and provide usable code
or libraries one can link to as a physisist: great!

The real problem is that most of teh code written by physicists is hard
to maintain: I think XLoops suffered from Maple-changes in the past.
I and Frederik (while not actively doing research in physics any more)
try to fix bugs and adapt FeynCalc and Phi to
each new Mathematica version: already this is a pain. Unfortunately
there is no Computer-Algebra-IDE, no refactoring,
no established way of writing Unit-Tests (test-suites), etc. etc.
And physicists are not good in providing their code to the public (which
is not always on bad intention,but because one
might be ashamed of it (I once were of some code), or does not want to
answer questions from people (Paolo Nogeira , who wrote
that utterly fast Feynman Diagram generator code years ago, was once in
this mood).
Jos Vermaseren, the undisputed leader in Feynman-diagram-programs, once
mentioned that he was thinking of putting
some version of FORM in the open source domain; already now you can just
download FORM from
http://www.nikhef.nl/~t68/
and use it scientifically.

BTW: In 1990, before making the final decision to switch from Macsyma to
Mathematica, I visited the IBM Reserach Labs with
a friend and we got a nice look a Scratchpad II ( which would later turn
into Axiom), but it was clear that mathematicians wrote that
program and that for research problems in physics Mathematica (as a
rewrite system with typeless constructs) was much more useful.

Regards,

Rolf Mertig

GluonVision GmbH
Berlin
Germany

>Greetings! There is considerable interest in the maxima community to
>get FeynCalc working with Maxima, the open source branch off of to
>formerly popular commercial product Macsyma. We feel this would
>extend the availability of your fine work considerably, even making it
>possible to distribute with the popular GNU/Linux distributions all
>over the world. Is there any interest in the FeynCalc community to
>help with such an effort? If so, it would probably be helpful for
>such interested persons to identify themselves and their area of
>expertise. Also, suggestions for the simplest way to proceed would be
>helpful. Right now our best option appears to be to start with a
>rudimentary mathematica to maxima translator called mockmma. I don’t
>suppose that mathematica happens to compile its code to lisp at an
>intermediary stage which can then be inspected by the user?

>Take care,

James Amundson <[amundson_at_HIDDEN-E-MAIL]
<amundson_at_HIDDEN-E-MAIL\&replyto=[54pt9gfhbe.fsf_at_HIDDEN-E-MAIL]\&subject=Re:%20%5BMaxima%5D%20status%20of%20gamalg\&from=>>
writes:

/> On Mon, 2004-05-03 at 16:26, Richard Fateman wrote: /
/> /
/> > I think that if FeynCalc is exactly the specification /
/> > you want, that is useful information: it could be /
/> > possible to replicate it in Maxima. I don’t know how /
/> > many people know both Mathematica and Maxima, but /
/> > you are clearly one of them! /
/> /
/> I would strongly encourage someone to look at getting FeynCalc to work /
/> with Maxima. They FeynCalc authors may (or may not) be interested in /
/> cooperating. It would be worth asking them. /
/> /
/> I owned a copy of FeynCalc back when it was a commercial product. I /
/> don’t have time to work on it right now myself, but I can promise
that I /
/> will be cooperative with someone who does work on this project. /
/> /
/> I think investigating using MockMMA to help would be an excellent /
/> research project. /
/> /
/> –Jim /
/> /
/> _______________________________________________ /
/> Maxima mailing list /
/> [Maxima_at_HIDDEN-E-MAIL]
<Maxima_at_HIDDEN-E-MAIL\&replyto=[54pt9gfhbe.fsf_at_HIDDEN-E-MAIL]\&subject=Re:%20%5BMaxima%5D%20status%20of%20gamalg\&from=>
/
/> http://www.math.utexas.edu/mailman/listinfo/maxima /
/> /
/> /
/> /