Re: Puncher soft limiter !!!

[Lorenzo Fiorentini]

If you first check the argument passed, you will find theta the pbm is not 
in your function, but in the way Octave (and matlab) computes the last 
element of your [a:step:b] argument to the function.

Not always the last element of your vector [a:(b-a)/n:b] is b. Sometimes it 
is b-(b-a)/n, because of rounding done n times when adding (b-a)/n.
NB. If the step is not a power of two [2^(-k)], there must be some rounding 
on a binary machine.

Try adding eps or 2*eps or step/10 to b, or use a different sintax like 
(a+[0:n]*(b-a)/n).

I am (pretty ;) sure it will work.

Lorenzo

On Wed, 4 Jun 2003 01:57:10 +0100 (BST), cctsim <address@hidden> wrote:

> Hi all,
>
> I have written the c++ code below to implement a
> puncher funtion for a soft limiter. By definition, that is
>
> f(x)=x for -A<=x<=A and f(x)=0 otherwise
>
> I thought it was working fine until I noticed
> the following problem:
>
> > > max(puncher(5,5))
> ans = 5
> > > max(puncher([0:0.1:5],5))
> ans = 5
>
> which are ok, but
> > > max(puncher([-5:0.1:5],5))
> ans = 4.9000 which is wrong !!

--




-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------