[SciPy-user] neighbourhood of randomly scattered points

Robert Kern robert.kern@gmail....
Wed Aug 29 18:48:38 CDT 2007

fred wrote:
> Robert Kern a écrit :
>> White noise isn't all that uniform. It does clump, and some kinds of displays
>> may show that off more than others. The orthogonal striations are a side effect
>> of your square definition of "neighborhood" that you used to plot and not
>> anything intrinsic to the data.
>> If you do need something more spatially uniform than what real pseudorandomness
>> gives you, then you should look at low-discrepancy sequences.
>>   http://en.wikipedia.org/wiki/Low-discrepancy_sequence
> Sorry, I don't understand.
> 1) My algorithm to enumerate the number of neighbours in the 
> neighbourhood does work fine
> and has been validated.
> 2) http://fredantispam.free.fr/rect.png
> shows the result for a rectangular neighbourhood. That's ok,
> there are striations, but this is not the problem.
> 3) http://fredantispam.free.fr/circ.png
> shows the result for a circular neighbourhood. There is no more striations.
> The question is : why I get (for the circular case) around 90 neighbours
> at il = xl = 450 and around 40 at il = xl = 200, ie less than half, if 
> it is
> completely random ?

Show us your code, if you think there is a problem in it. Looking at images is a
very poor way to judge randomness; intuition isn't very good. I'm trying to
verify your numbers quantitatively, but without knowing how you implemented the
circular neighborhood, I can't. If you picked the most dense spot and the least
dense, then those values appear reasonable.

Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
 that is made terrible by our own mad attempt to interpret it as though it had
 an underlying truth."
  -- Umberto Eco

More information about the SciPy-user mailing list