Re: bad random numbers

[heberf]

I don't think this needs to be a concern.  Consider the following 

octave:1> z = zeros(500,1);
octave:2> Y = X = zeros(100,3);
octave:3> for m = 1:500
> for k = 1:1000
> Y(k,:) = rand(1,3);
> X(k,:) = rand(1,3);
> endfor
> M = cov(X,Y);
> z(m) = M(1,1);
> endfor

This takes a while but if you look at z you will see that it takes on positive 
and negative values is centered around zero as it should be.  So the first 
elements of two successive rand(1,3) operations are uncorrelated.

Heber Farnsworth

Gerald Jansen wrote: 
> "John W. Eaton" wrote:
> > On 13-Jul-1999, Mike Miller <address@hidden> wrote:
> > | Note that the resulting matrix always had a very small number in the
> > | 1,1 cell.  The value in that cell ranged from .003 to .017 ...
> >
> > The initial seeds for rand are computed from time().  Not random
> > itself, but at least it should be different each time you run Octave.
> 
> Perhaps the first random number should be discarded if the seed computed
> from time() leads to patterns, like those observed by Mike, on some
> systems.
> 
> -- 
> Gerald Jansen
> 
> 
> 
> ---------------------------------------------------------------------
> Octave is freely available under the terms of the GNU GPL.  To ensure
> that development continues, see www.che.wisc.edu/octave/giftform.html
> Instructions for unsubscribing: www.che.wisc.edu/octave/archive.html
> ---------------------------------------------------------------------
> 



---------------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.  To ensure
that development continues, see www.che.wisc.edu/octave/giftform.html
Instructions for unsubscribing: www.che.wisc.edu/octave/archive.html
---------------------------------------------------------------------