Re: Arguments for supporting Motion P1788/0023.01:NoMidRad
Rudnei, P1788
On 15 Sep 2010, at 01:46, Rudnei Cunha wrote:
> I strongly agree with Nate. I've seen enough evidence in the field of numerical linear algebra using interval arithmetic - both standard and mid-rad representations - that have convinced me that, in this field at least, mid-rad is the best choice. Leaving it as a sort of "outcast interval arithmetic" would not be wise for the scientific community, specially considering that numerical linear algebra is at the core of large-scale scientific applications.
Can you point us to just one article where the advantage of a mid-rad type in interval numerical linear algebra is demonstrated, with algorithm(s) and performance comparisons? Our standard is about _rigorous_ enclosures, so it must be about those, not approximate ones. And INTLAB's fast matrix multiplication won't do, because it doesn't use a mid-rad type, only a split into a "mid" and a "rad" point matrix.
John