[clean-list] Re: Functions that look the same

[Siegfried Gonzi]

clean-list-admin@cs.kun.nl wrote:

> From: Arjen <arjenw@sci.kun.nl>
> To: smithll <smithll@cpsc.ucalgary.ca>
>
>
> Overall, one should be aware of these imprecision issues when using
> floating point calculations. I recommend you use (abs x) < epsilon
> with epsilon sufficiently small instead of x == 0, regardless of which
> programming language you use.

There are articles (reliable) adressing that topic:

- "What every computer scientists should know about floating-point
arithmetic":

http://docs.sun.com/htmlcoll/coll.648.2/iso-8859-1/NUMCOMPGD/ncg_goldberg.html

- or Kahan's site:

http://www.cs.berkeley.edu/~wkahan/

and the articles therein (especially of interest: "How Java's floating-points
hurts everyone everywhere")


S. Gonzi