[Axiom-mail] Re: reclos.spad

[root]

> Renaud Rioboo wrote:

> > I just saw you mention reclos.spad on sci.math.symbolic.
> > Can you point me at a reference for the reclos.spad code? 
> > I'd like to track down the author and get permission to use
> > quotes from the papers in the literate document for that domain.
> 
> I am the author of reclos.spad. It is part of Axiom, you should have it since
> Mike Dewar asked me for copyright this summer. I saw him in spetember and he
> told me everything was OK.
> 
> I wanted to get in touch with you because I am finishing version 2. I was
> wondering wether it could be part of your new distribution.
> 
> Algorithms underlying release one are described in
> 
>           ``Computation Of The Real Closure of an Ordered Field'' full paper à
>      ISSAC'92 (International Symposium on Symbolic and Algebraic
>      Computations). ACM Press, SanFrancisco (1992).
> 
> The effective general closure model is described in
> 
>           ``Generic Closure Of an Ordered Field, Implementation in Axiom'' en
>      commun avec Z. Ligatsikas et M. F. Roy. Mathematics and Computers in
>      Simulation Volume 42, 1996. Présenté à IMACS'93 (International
>      Mathematics and Applications to Computer Science)
> 
> Algorithms for version 2 are in
> 
>           ``Polynomial Gcd Computations over Towers of Algebraic Extensions''
>      en commun avec Marc Moreno Maza. Lecture Notes in Computer Science 948,
>      Springer Verlag. Présenté à AAECC'95 (Algebraic Algorithms an Error
>      Correcting Codes).
> 
> and
> 
>           ``Towards Faster Real Algebraic Numbers''. Présenté à ISSAC'02
>      (International Symposium on Symbolic and Algebraic Computations).
> 
> 
> In case you don't have them, sources for RECLOS version 1 are at
> 
> http://ftp.lip6.fr/lip6/softs/formel/Axiom/RealClosures
> 
> you may substitute ftp:// for http:// and acces it using a traditional ftp
> client. I'll put version 2 there too.
> 
> Let me know if you want the papers. There might be other things from my home
> page 
> 
> http://www-calfor.lip6.fr/~rr/
> 

Thanks for the references. I'll look at them this weekend.

Yes, I have the code for version 1. I'll update the code to use Version 2
if you'll put it someplace or send it to me. I've already received a couple
fixes to the algebra tree that will be distributed with the next version.

I've signed you up for both the address@hidden and the
address@hidden mailing lists. The archives for these
mailing lists can be reached from the website. 

Since you haven't been on the mailing lists you will have missed my
discussions of literate programming and Axiom. I'm modifying the whole
system to use a literate programming style. All of the files in the
distribution will be "pamphlets" or "booklets". 

The literate programming idea comes from Knuth and Dijkstra. Knuth
generated a language called Web which takes a "literate document"
and has two functions, tangle and weave. Tangle generates code, 
weave generates tex thus:

   Literate Document
                     -> tangle -> pascal code 
                     -> weave  -> .tex file

Norman Ramsey has noweb which is not language specific so the process is:

   Literate Document
                     -> notangle -> any code 
                     -> noweave  -> .tex file

A pamphlet is a file is basically a tex document with a couple extra
tags to mark the "code chunks". Basically you write the tex document
and when you want to introduce code you wrap it with:
<<(some string)>>=
   your source code
@

These code chunks are extracted by notangle. 

In the new Axiom all of the makefile, C, lisp, boot, input and algebra
files are now (or will shortly be) in pamphlet form. Thus reclos.spad
will now be reclos.spad.pamphlet and the makefile will do:

   reclos.spad.pamphlet
                     -> notangle -> reclos.spad -> compile
                     -> noweave  -> reclos.tex  -> reclos.dvi

At the present time the reclos.pamphlet file contains only the source code
so the reclos.tex file also only contains the source code. What I'd like to 
do is use the original documents to explain the code. 

The motivation for this is that the theory is currently enshrined in a 
library and the code is in the running system. In order for the next
generation of people to be able to understand, maintain and update the
algebra code we need to bring these two things together. So the idea
is to embed the original tex documents into the reclos.spad.pamphlet
file along with the code, modify and rewrite the document so that the
gap between the theory and the code is covered and add the resulting
pamphlet to the system. I'd like to see this happen with reclos.spad.
In order to do that I'd need your permission (and review) to use your
papers in the documentation of the code. Is this possible?

Booklets, mentioned above, are a meta-level to pamphlets. The idea
is that a booklet represents either a horizontal slice of the system
(e.g. all matrix functions) or a vertical slice (e.g. integration from
top to bottom). Booklets would also be useful for teaching (e.g.
linear algebra). 

Even more interesting is that I've found out that IBM Research did an
interactive textbook based on Axiom. Commercially this would be dead,
of course. I've been trying to contact Rudiger Gebauer, who, I've been
told, was the publishers contact with IBM. I'm hoping that they would
be willing to open source the textbook which would give the whole
community a great benefit. We'd have a good textbook and a working
first try at interactive textbooks. I'm sure that the CAS community
can take it even further given the chance.

> I can send you a copy of my habilitation which explains a little more about
> real algebraic numbers and discusses semantics for computer algebra programs
> if you want. I will defend it on December 18.

Please do send it.

> I saw your intervention on the OSCAS mailing list about an Axiom journal. I
> went once to the home of the development for your new Axiom but there was not
> much by that time. My browser has problems with savannah's pages which are
> HTTP-1.1 specific. You can add me to the different lists there.

You've been added to both lists. OSCAS isn't Axiom specific so I've 
tried to limit my cross-postings on that list. On the other hand, OSCAS
has been dormant for a few months and I'm not sure what the current state
of the community is. There was talk at one point of adopting Axiom but
I've heard no further discussion on the subject. I tried to contact
Gilles Khan but email to his address bounced. Has he left his position?

> By the way, could you advance on a Bill Shelter award ?

Nothing much has taken place on the Schelter award. I did nominate
Bill for the Free Software Foundation award. I had a discussion with
Richard Stallman about the potential to combine the FSF award and
the Schelter award idea. It seems reasonable to me but I'm not sure
if everyone associated with the FSF award would agree. In any case,
it looks like I'm going to end up putting up my own cash in order
to fund the award. I'd also need to get committee volunteers so the
award could be decided and the cash distributed by someone other than me.
Depending on my financial circumstances I've been looking at offering
a $500US award at IJCAI next year. Other than that there is no progress
to report.

Tim