Re: More on trits & tetrits... (long)
Vincent Lefèvre wrote:
> In floating-point arithmetic, if you have an exact zero and signed
> inexact zeros, you need a 4th zero (inexact zero with unknown sign)
> and a 3rd infinity (infinity with unknown sign).
Right -- except it would be a 5th infinity. (That's all in staying
with the affine interpretation. In the projective interpretation
there would be only two zeros (exact and inexact) and two infinities
(Overflow and Pole), both without a sign. Early plans for 754 had
not yet settled on affine vs projective, and the Intel 8087 math
coprocessor supported both (1984).
Now we just need a (paper/dream) arithmetic design for this!
Interval Arithmetic is supposed to get us out of this mess.
Michel
---Sent: 2010-04-23 14:32:41 UTC