[IPOL discuss] Remarks and tips about numeric precision

Nicolas Limare nicolas.limare at cmla.ens-cachan.fr
Mon Dec 16 04:44:13 CET 2013

On Sat, Dec 14, 2013 at 04:19:03PM +0100, Miguel Colom wrote:
> Some useful short remarks and tips about numeric precision:
> http://www.codeproject.com/Articles/29637/Five-Tips-for-Floating-Point-Programming

On the first point, how to compare floating-point numbers, and what is
a good epsilon for this comparison, one good reference is here:
-> http://randomascii.wordpress.com/2012/02/25/comparing-floating-point-numbers-2012-edition/

It's an updated version of the previously written page, which I have
seen mentioned as the reference in many places:
-> http://www.cygnus-software.com/papers/comparingfloats/comparingfloats.htm

Basically, the idea is to use an epsilon definition (ULP) depending on
the precision of the floats, and to do that efficiently by comparing
floats like integers!

The same author has written many other things about floating-points,
all very useful and well detailed.
-> http://randomascii.wordpress.com/category/floating-point/

And there is the classic "What Every Computer Scientist Should Know
About Floating-Point Arithmetic" by David Goldberg, 20 years ago:
-> http://docs.oracle.com/cd/E19957-01/806-3568/ncg_goldberg.html

Nicolas LIMARE - CMLA - ENS Cachan        http://limare.perso.math.cnrs.fr/
IPOL journal                                            http://www.ipol.im/
-> image processing, reproducible research, open science
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <https://tools.ipol.im/mailman/archive/discuss/attachments/20131216/fd001064/attachment.pgp>

More information about the discuss mailing list