On asymptotically optimal estimates for general observations (Q1965884)

The authors consider asymptotically maximum likehood estimators and estimators asymptotically minimizing criterial functions of observations in statistical models with generalized sequences of observations. New necessary and sufficient conditions for the consistency of these estimators are established. The use of these conditions is illustrated on regression models with Gaussian and contaminated observations as well as on models of exponentially distributed random processes and fields.