## Section: New Results

### Floating-Point Arithmetic

Participants : Guillaume Hanrot, Philippe Théveny, Paul Zimmermann.

Together with Siegfried Rump, Sylvie Boldo and Guillaume Melquiond,
P. Zimmermann published a new efficient algorithm to compute the
predecessor or successor of a floating-point number in rounding to
nearest mode [7] . This algorithm is
about two times faster than the `nextafter` function from the GNU C
library.

With Vincent Lefèvre from the Arenaire project-team, and Kaveh Ghazi
(GCC developer), Ph. Théveny and P. Zimmermann submitted an article
entitled *Why and how to use arbitrary precision* to the journal
*Computing in Science and Engineering* .