Mini-Sized

Could a kd-tree be made less than one thousand bytes?

I'm not sure, if you leave out the complicated stuff it loses its speed. Perhaps with the help of built-in structures like Java's PriorityQueue it might be possible. Impress us!

I'd say... possibly, but there's no reason to. An efficiently implemented linear search will waste much less of the limited codesize, and ought to be fast enough, espescially given that the number of search dimensions that make sense to put in a Minibot is small. The first nearest-n-neighbor search bots didn't use kd-trees and worked just fine.

On the other hand... if you're wanting to do it for it's own sake, it would be real neat to see just how small it could get! :)

I will be rather busy over the next fortnight, but after that I could give it a try.

Do you, Rednaxela, know of any efficient, low-codesize linear search algorithms with multiple choice?

I ran across this while doing my KD-Tree research http://www.michaelpollmeier.com/selecting-top-k-items-from-a-list-efficiently-in-java-groovy/ It looks like once you calculate your distances to everything (two for loops) it's just about 4 lines of code.