Re: [Texmacs-dev] Math input / output from CAS (Axiom)

[Joris van der Hoeven]

On Wed, Nov 30, 2005 at 10:22:17PM -0800, Karl Hegbloom wrote:
> When I use mathematical input to an Axiom session from within TeXmacs, I
> can't type the == for delayed assignment. It gets turned into the long
> equal sign. That's alright, perhaps, as long as the right thing gets
> sent to the Axiom process.

I fixed this issue.

> When Axiom outputs a matrix, it outputs a table wrapped in large [],
> rather than a \matrix object.  So, when I paste that into a document,
> then later on go to paste it back into an Axiom session, it's not a
> matrix.  Try this:
> 
>      1. Open a new document in TeXmacs, and in there, open an \eqnarray*
>         or \[ and enter a \matrix with a few columns and rows filled in.
>      2. Start an Axiom session, then highlight and paste the matrix into
>         the session.
>      3. Take the output of that, highlight everything including the
>         square brackets, and paste it into the next input line of the
>         Axiom session. The result is not a matrix, but a List Matrix ?.
> 
> What's happening is that tables are seen as matrices. I suppose that's
> alright, but maybe a table wrapped with large [] should have the []
> silently stripped by the input stream filter.  The thing is that in
> TeXmacs, there's a different object for a determinant and for a matrice.
> Both are defined in mth-std.ts, and internally contain tables.  It just
> seems like the output from the CAS should be a \matrix rather than a
> \tabular.

Yes, this should be fixed by the people who wrote the axiom interface.
Notice that TeXmacs now supports more and more semantic markup,
although this is not yet documented. See the "tmprint1" routine
in the interface with Maple. It would be good to gradually move
the LaTeX output of most current interface to this semantic output:
this will allow us to implement additional features soon.
Moreover, semantic markup is much easier to produce than
good LaTeX output (as I experienced with the Maple plug-in).

> It would also be nice to have automatic coercions from vector to column
> matrix.
> 
> Another annoyance is that the Axiom output does not have the *
> explicitly written into polynomials. Instead, the coefficient and
> variable are juxaposed, and so they are typeset too close together, and
> when you paste the expression back into Axiom, it doesn't work right.
> Sometimes, I've noticed, it appears to do something, but gives incorrect
> results. Usually, it signals an error about not having an operator
> defined for the variable name.

Again a problem with the output converter.

> Ideally, there would be a better communication format between Axiom and
> TeXmacs.  LaTeX / TeX are not really designed to convey the semantics of
> the expressions, only the appearance, isn't that right?  There's a lot
> of research out there on this subject...

Heh ^^^.