On 3/9/07, Raymond Toy <raymond.toy at ericsson.com> wrote:
>
> >>>>> "Stavros" == Stavros Macrakis <macrakis at alum.mit.edu> writes:
>
> Stavros> an INF in the system, you are committing yourself to losing
> information
> The current code does have 1/(2*inf) = 1/(3*inf) = 0. But why is
> 1/(1/2*inf) = und? Why is this not 0?
Oops. I completely messed up what I wanted to say. Let me start again with
a clearer, simpler example:
I think we all agree that 2*inf = 3*inf = 0, and that inf/inf = (2*inf)/inf
= (3*inf)/inf = UND. That is, the simplifications involving inf lose
information about where the inf came from, just like simplifications
involving 0: 2*0=3*0, and 0/0 = (2*0)/0 = (3*0)/0 = UND, not 1, 2, and 3.
The point of all this is that once you inject Inf into your number system,
you start getting some surprising results (like 1^inf = und) because your
expectations are based on axioms which are no longer valid.
-s