EM Algorithm for Estimating k-Dimensional Gaussian Mixture Models with Implementation Details

Implementation of the Expectation-Maximization algorithm for estimating k-dimensional Gaussian mixture models. The algorithm accepts input data matrix X(n,d) with n observations and d-dimensional variables, maximum Gaussian components k, likelihood tolerance ltol, maximum iterations maxiter, plotting flag pflag, and initial parameter structure for weights, means, and covariances. Returns estimated mixture parameters and log-likelihood value.