Statistical Inference with constraints aims to design procedures that, taking advantage of the a priori information that is often available on the parameters of the model of interest, are more efficient than the classic ones that do not take into account such prior information.

Since its inception more than twenty years ago, this group has been dedicated to the study of general estimation and hypothesis testing problems in normal linear models when the incorporation of a priori information results in the reduction of the usual parametric space to a cone. convex and closed.

The methodology has been based on the principle of maximum likelihood and, consequently, on the study of the behavior of the maximum likelihood estimator and the likelihood ratio test.

The main contributions of the group are, on the one hand, to delimit in which situations the procedures with restrictions have good behavior and to highlight and explain the structural reasons that justify the poor behavior of these procedures in other situations, and, on the other, in the proposal of alternative procedures that avoid said anomalous behavior.