Using MPI to implement parallel kmeans algorithm. It means that this is a parallel implementation that you can you can deal with much larger data set.

src : c source code is in this directory.
visual : every iteration, the middle result is output in this directory, then use these output to do visiualization using python matplotlib.
python : python code to do visulization using matplotlib.

In this experiment, I generated centroids for three clusters. Then randomly generated 500 points, each
point randly belog to one cluster, then generate a distance to this centroid.
Using the implemented mpi-kmeans to cluster these data, for every iteration, save the out of kmeans as a png file. The result is showd below.