On asymptotic normality of the least square estimators of an infinite- dimensional parameter (Q1332013): Difference between revisions

The asymptotic properties of the least squares estimators (LSE) of nonlinear regression parameters are studied in the case when these parameters belong to \(\Theta\), a subset of an infinite-dimensional space. For \(n\) observations the random functional \(Q_ n (\theta)\) and the corresponding LSE \({\widehat {\theta}}_ n\) are defined. Let \(\Theta\) be a convex closed bounded subset of a separable, reflexive Banach space and the functional \(Q_ n\) satisfy a continuity condition that is stronger than the continuity in norm. Then for LSE's the consistency of functionals of estimators, the asymptotic normality of some functionals and the weak convergence of normalized estimators to a Gaussian distribution are proved.