Uniform convergence and rate adaptive estimation of convex functions via constrained optimization (Q2862446)

Convex estimators and B-spline estimators are constructed which are either adaptive with respect to the Hölder exponent \(r\) or are chosen for such a fixed exponent. The background for this work lies in the interest in asymptotic analysis and adaptive convex estimators for the aforementioned Hölder space using uniform norms. Splines and concretely B-splines turn out to be most useful for this paper. Indeed, for the B-spline case optimal rates of convergence are established. For these purposes, uniform Lipschitz properties in the uniform norm of optimal B-spline coefficients are derived in advance, as well as a formulation of the estimator and its coefficients in terms of constrained optimisation. A number of numerical examples are included at the end to illustrate the results.