[Richard Fateman <fateman@cs.berkeley.edu>, Wed, 08 Sep 2004 08:33:34 -0700]:
> You can define such objects, but CAS generally don't
> "understand" them particularly well, including Mathematica.
> There's a KDELTA for Kronecker Delta in the tensor package.
> But integration and simplification need to smarten up
> to work with such things.
Thanks for the reply.
For the application at hand I just look whether ev(is(r > 0), numer,
prederror: false) happens to return true or false, and error out
otherwise, which is sufficient for now.
I only wanted to make sure that I am not missing something more
elaborate where someone already thought about all those tricky
questions that you mention (and that I was well aware of). In
Mma-terms, I wanted to avoid using the newbie's attempt of If[# > 0,
1, 0, Throw[]]& when UnitStep has already been implemented.
Albert.