From: Marcin Raczkowski <mailing.mr@...>
Date: 2008-01-21T09:36:35+09:00
Subject: Re: kD-Trees implementation

>> about finding median - my algorithm doesn't find it either
> 
> But I think it does - you sort N elements and use the N/2'th
> element, which is the median. That guarantees a balanced tree.
> 
Well, I guess it's close enought - it doesn't guarantee balanced tree - 
you have to take N+1 cases into consideration.


> The r-tree (and priority r-tree) looks like a useful structure
> too - the equivalent of a b-tree for spatial data.
> 
> Clifford Heath.

Well I started implementing it as brain-tease, and platform to test my 
optimization techniques - so plan is to implement full set of operations 
- and then start to optimize it as much as i can ;)

One of properties of kd-trees is important to me - fast finding 
neighbours in k-dimensions