Re: [Algorithms] Converting floats to a specific range

[Tom F. <tom...@ee...>]

>    x - floor(x)
> 
> which is always nonnegative.

Obscure fact - I *think* it is always non-negative, but it might not be less
than 1.0. For example, if x is very small and negative, for example x=-1e-10
then floor(x) = -1.0f (correct), and then -1e-10 + 1.0f = 1.0f. It's
possible you may be able to change your math rounding mode to "round down"
and then it will return the smallest number less than 1.0f - I've never
tried it.

But yeah, frac(x) can counter-intuitively return 1.0f sometimes.

TomF.


> -----Original Message-----
> From: Fabian Giesen [mailto:ry...@gm...]
> Sent: Friday, May 07, 2010 4:56 AM
> To: Game Development Algorithms
> Subject: Re: [Algorithms] Converting floats to a specific range
> 
> On 07.05.2010 13:36, Zafar Qamar wrote:
> > Hi,
> >
> > I'm sure this problem is pretty simple, but here goes...
> >
> > I have a set of float numbers that I'm trying to auto-wrap within a
> specific range of 0.0 to wrapSize (wrapSize is positive, and not a
> power of 2)
> >
> > So, if I choose a wrapSize of 1.0 then I want the following to happen
> when I feed in numbers...
>  > [..]
> 
> There's a standard C library function, fmod, that nearly does what you
> want. The one thing you need to fix is that fmod(x, wrapSize) is
> negative whenever x is. So the solution to your problem is
> 
>    fmod(x, wrapSize) + (x < 0.0 ? wrapSize : 0.0);
> 
> The special case for wrapSize == 1 is quite easy, and worth
> remembering,
> since you get to avoid the conditional: You can just compute
> 
>    x - floor(x)
> 
> which is always nonnegative.
> 
> If your target platform/language doesn't have fmod but does have floor,
> you can usually make do with
> 
>    x - wrapSize * floor(x / wrapSize)
> 
> but if you use this you need to watch out if you might encounter very
> large values of x in your application - as soon as floor(x / wrapSize)
> gets to the range where neighboring floats are more than 1 apart, this
> expression will break.
> 
> -Fabian
> 
> -----------------------------------------------------------------------
> -------
> 
> _______________________________________________
> GDAlgorithms-list mailing list
> GDA...@li...
> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
> Archives:
> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-
> list
> No virus found in this incoming message.
> Checked by AVG - www.avg.com
> Version: 8.5.437 / Virus Database: 271.1.1/2838 - Release Date:
> 05/06/10 18:26:00