-#' @return a list with several models, defined by phi, rho, pi, llh
+#' @return a list with several models, defined by phi (the regression parameter reparametrized),
+#' rho (the covariance parameter reparametrized), pi (the proportion parameter is the mixture model), llh
+#' (the value of the loglikelihood function for this estimator on the training dataset). The list is given
+#' for several levels of sparsity, given by several regularization parameters computed automatically,
+#' and several ranks (between rank.min and rank.max).