Re: Motion 19.02 NO
On 2010-11-02 05:09:43 -0500, Ralph Baker Kearfott wrote:
> Vincent,
>
> On 11/2/2010 04:28, Vincent Lefevre wrote:
> >Users of mid-rad may prefer performance over reproducibility.
> >
> Yes. However, a prominent person in the general scientific computing
> community has expressed the importance of reproducibility to me, and
> those arguments seem reasonable.
Reproducibility is a nice property, but correctness[*] (implied by
the FTIA) can be sufficient.
[*] something one doesn't have with floating-point arithmetic, except
for applications where all the computations are exact (testing the
inexact flag would prove the result), but this is very rare.
--
Vincent Lefèvre <vincent@xxxxxxxxxx> - Web: <http://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / Arénaire project (LIP, ENS-Lyon)