Operations on inf



On 3/10/07, Richard Fateman <fateman at cs.berkeley.edu> wrote:
> I've been away for a week, but inf is "undirected infinity" then are the two
> Symbols, each  inf,  the SAME?   If not, then inf*inf might be -inf.

I believe Maxima consistently (for once) treats INF as positive
infinity: limit(1/x,x,0) => und (not inf).

> Like intervals, [-1,1] * [-1,1] ... Is it the same interval or two different
> objects that happen to have the same description.

I believe that the consensus is that each inf has to be treated as
independent, thus inf/inf => und, not 1 or ind.  Obviously this is a
pessimistic assumption, but since we currently don't carry around the
additional information necessary to do better....

> some program returns "undefined"  two  times, and subtracts one from the other to get  0.

Yes, that is one motivation for making sure that und-und => und, etc.
Currently, for example, integrate(sin(x)-cos(x),x,0,inf) => 0 (!!!).

             -s