Multi-dimensional nearest-neighbor code

[Lance Norskog <go...@gmail.com>]

Hi-

I'm working on an N-dimensional grid-based hashing mechanism, and would 
like a large N-dimensional dataset (real-world or generated) for testing 
(and showing off :). This would be 100s or 1000s of dimensions, which 
fit into at most 20G of memory. Sparse is preferred.

I would like to go head-to-head with other algorithms. So if there's a 
zillion-dimension variant of something in the current Mahout base, that 
would be great.

Lance Norskog