On 3/18/08, Shahir Molaei <shahir at inbox.com> wrote:
> I have written such a code in Mathematica for definite n and
> I 'know' what the pattern of the eventual outcome will be for that
> block and for general n. However computations take too much time
> as n exceeds 3 (n is greater than 2, this is a sort of constraint) and
> still they are for specific n. This is mainly because the metric is
> itself complicated. Strictly speaking, I want to use Maxima as a
> means of providing the proof for a form I know in advance.
This sounds like a very interesting research program, but
unfortunately I do not believe Maxima can help much with that.
Maxima does not have a built-in indefinite matrix type,
although it does have the machinery that makes it
possible to define such a type. I don't know about Macsyma's
indefinite matrix; probably it was added post-1982 (the year
of the "DOE Macsyma" fork which is the ancestor of Maxima).
I would imagine that Maxima's machinery for new types is
comparable to Mathematica's, so if you are already working
in Mathematica maybe that is the way to go.
Incidentally someone is still selling commercial Macsyma from
a web site; maybe it is worth it to go ahead & buy a copy.
Sorry I can't be more helpful.
Robert Dodier