[SciPy-user] remove duplicate points...

[Nathan Bell]

On Wed, Apr 23, 2008 at 10:50 AM, Angus McMorland <amcmorl at gmail.com> wrote:
>
>  As a quick test:
>
>  sz = 1e7
>  sz - np.unique(np.random.normal(size=(sz,3)).prod(axis=1)).size
>
>  returns 0 - i.e. no common products unless the points are deliberately
>  made the same. Am I mis-thinking this?

I think so.  You example shows that 2^53 is a much bigger number than 10^7.

OTOH the likelihood that two different points have a zero coordinate is higher.


-- 
Nathan Bell wnbell at gmail.com
http://graphics.cs.uiuc.edu/~wnbell/