Title: When do discretized data have a good behavior in functional nonparametric regression?

Abstract: We consider two families of nonparametric regression estimates in the context of Functional Data Analysis (FDA). We study the relationship between the consistency of the estimators when using the full curves and the consistency when only a discretization of the data is available. As a consequence, we derive the consistency and rates of convergence for most of the regularization techniques used in FDA, including smoothing, interpolation and basis representation.