Extended real arithmetic (was Re: inf - inf = 0 ??)



  At the risk of repeating myself, I can refer you to a paper that 
suggests this fits within
a framework of interval arithmetic.
e.g.  interval(-inf, inf) means  ---  there is a real  number but we 
don't know anything about its value.
this is perhaps indefinite (IND)

an empty interval  (we can choose one to be canonical) that has nothing 
in its interior means
--- there is no number that this could be, which is perhaps undefined (UND)

http://www.cs.berkeley.edu/~fateman/papers/interval.pdf

There are what I consider misuses of intervals regarding limits that are 
present in Mathematica and Maple,
eg..

limit(sin(x),x,inf) ?=?  interval(-1,1).

This leads to either erroneous answers to limit problems and things 
related to limits, or to a substantially
overloaded notion of "equality".   Like any interval that contains 0  is 
equal to 0.

Whether this can be put together in a computer algebra system to be more 
consistent is possible, but
consistency in general is tough, and certainly not achieved now.
There are many examples in which, for some procedure f,

f(a)  is   different from  subst(a,x,   simplify(f(x)) )
where x is some arbitrary symbol and   a   is some particular value.


RJF