Explore chapters and articles related to this topic
Variable Selection for the Prediction of C[0,1]-Valued Autoregressive Processes using Reproducing Kernel Hilbert Spaces
Published in Technometrics, 2019
Beatriz Bueno-Larraz, Johannes Klepsch
By Lemma 1 of Pumo (1998) we know that for some positive constants A1, A2, A3,
where m is the sample size and 0 < r < 1 is given by assumption H1. By Borel–Cantelli, if the series in m whose terms are these probabilities is convergent for every ϵ > 0, we get the almost sure convergence stated in Equation (12). The sum is of order
where Cϵ, Cr, Dϵ > 0 and these three series converge (for example, by the limit comparison test with ∑m− γ, γ > 1). Concerning (13), the same Lemma 1 of Pumo (1998) states that the bounds for these probabilities are equivalent but with m − 1 in place of m.