Re: de-for-ing..

[David Bateman]

Miroslaw Kwasniak wrote:

> On Sun, Feb 06, 2005 at 10:42:36PM +0100, Gorazd Brumen wrote:
>  
>
> > The results are the same, but the for loop is much
> > faster and less memory consuming. the kron solution
> > is preety much stupid.
> >    
>
> I'm suprized too, because octave has slow loops.
>
>  
>
> > Is there a better way than the for loop?
> >    
>
> I've tested also two additional methods with repmat: and sort:
>
> place3=sum(repmat(x_init(:),1,lx) <= repmat(x(:)',n,1));
> t3=cputime;
>
> and sort:
>
> [y i0]=sort([x_init, x]);
> i=find(i0>n);
> place4(i0(i)-n)=cumsum([i(1) diff(i)]-1);
> t4=cputime;
>
> Results for
> ftest(n=1000, lx=10000);
> ftest(n=10000, lx=1000);
>
> cpu-tme in seconds (2.1.63 on A1600XP)
>  for-loop   kronecker  repmat     sort
>  1.600000   2.350000   4.200000   0.040000
>  0.8700000  2.3800000  2.5200000  0.0100000
>
> Winner is the method with sort, but it may fail, because it assumes that
> sort doesn't change order for equal values:
>
>   x(i) == x(j) & i < j ==> find(i0==i) < find(i0==j)
>
> and it may depend on sort implementation.
>  
sort maintains ordering and so won't swap equal values...

D.

--
David Bateman                                address@hidden
Motorola Labs - Paris                        +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 1 69 35 77 01 (Fax) 
91193 Gif-Sur-Yvette FRANCE

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary



-------------------------------------------------------------
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
-------------------------------------------------------------