I\'m trying to create a GMM with a fixed variance for all features AND with that variance fixed across all cluster (i.e tied-spherical), is there an implementation for this? the