An implementation of the kmeans algorithm, using kdtrees to
speed up finding nearest centroids. There is both a CPU and
a GPU version (using CUDA). 

Build:
    $make

Run on randomly generated inputs:
    $./kdkmeans
To visualize the clustering for randomly generated inputs, run 
the show.sh script. This uses the temporary files generated by
kdkmeans in /tmp/ and visualizes the clustering.

Parameters (such as size of the problem) to the program can be
changed in common.h

You may want to change some parameters in the Makefile. For example,
it is set to build for sm13 by default. Also the -G flag is used.
(I faced cuda compiler issues without this flag).

NOTE:
I wrote this code before I started to use GIT, so there is nothing
much in terms of commit logs for this project.