maxima-bounces at math.utexas.edu wrote on 08/28/2008 09:01:21 AM:
> I'm not sure what is implemented, but the following sequence:
>
> x: inf;
> x-x;
> could result in zero, because it is the same infinity.
> so should (x^2-1)-(x+1)*(x-1).
>
> However,
> y:inf;
> x-y should result in undefined.
> Getting undefined arithmetic right would be just as important.
>
> Labeling each unique inf with an identifier is a way to do this.
No, my code doesn't tag infinities this way. If we simply tagged
each infinity with a gensym identifier, it wouldn't be all that
hard to do, but getting through the test suite might be a chore.
Tagging each infinity with a polynomial that includes its
complete history would be harder, I think.
Barton