Re: math function accuracy -- was Re: text2interval again /
On 2013-03-12 07:55:02 -0700, G. William (Bill) Walster wrote:
> With a containment-only standard, every implementations claim to be standard
> compliant can be falsified with just a single counter-example.
This is also true if some accuracy is required: you just a single
counter-example to falsify an implementation.
And it is not necessarily more difficult to find a counter-example.
Indeed even with a containment-only standard, an implementation can
choose to return the tightest interval, assuming the numerical library
is correctly rounded. But it may happen that the numerical library
assumes that there are no hard-to-round cases needing 3 times the
target precision. Said otherwise, finding a counter-example is more
or less equivalent to doing a hardest-to-round case search.
--
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 / AriC project (LIP, ENS-Lyon)