CHANGES in VERSION 1.8
----------------------

SIGNIFICANT USER-VISIBLE CHANGES
    o We now check if the covariance matrices can be inverted and if not we stop the algorithm and output NaN for the log likelihood. As a consequence, the function does not stop and can finish all computable values of K.

NEW FEATURES

    o Support for the multicore package. The number of cores will automatically be detected. The multicore package
    needs to be loaded for that.
    
    o Snowfall is also supported and if sfCluster() is called, snowfall will be used over multicore. So if you want to use snowfall make 
    sure not to load multicore
    
BUG FIXES

(start date: 10 July, 2010)
