On Mon, 27 Apr 2015 18:16:34 +0200 Ulrich Kulisch wrote:
The dot product is a fundamental operation in Numerical Analysis. It
allows a simple and very fast implementation on computers. *The exact
dot product* is the key operation for variable precision interval and
floating-point arithmetic. It would be a highlight of P1788.
Few people would disagree with the above.
As I see it, the issue is that simply including the exact dot product
would be woefully incomplete without also including a discussion of
the complete format as an arithmetic number format, and a couple of
other operations on complete numbers. If the only operation is to
convert a complete-format number to a floating-point number, then
we would have achieved nothing more than what is already required,
namely a correctly-rounded dot product.
Unfortunately the group did not want to include a whole chapter on
Complete Arithmetic, as that would have gone beyond the approved PAR.
(If I remember correctly, there was an earlier proposal in 2008 that
did include Complete Arithmetic.)
Michel.
---Sent: 2015-04-27 18:23:30 UTC