Bayesian Information-theoretic Criteria, for short BIC, is the widely used equation of selecting number of components of GMM. The equation is : BIC = $-2log(L) + klog(n)$ L is the likelihood. k is the component numbers and n is the sample size.