Estimates of the rate of approximation in the CLT for \(L_1\)-norm of density estimators (Q4473015)

Let \(X_1, X_2,\dots\) be a sequence of i.i.d. random variables in \(R\) with density \(f\) and \(\{h_n\}_{n\geq 1}\) be a sequence of positive constants such that \(h_n\to 0\) as \(n\to \infty.\) Then we may say that the author is interested in the classical kernel estimator \[ f_n(x)=\frac 1{nh_n}\sum_{i=1}^n K\left(\frac {x-X_i}{h_n}\right) \qquad (x\in R) \] for \(f\) and estimates the rate of strong approximation and probabilities of moderate deviations in the CLT for the \(L_1\)-norm of this kernel density estimator under no assumptions at all on the density \(f\) and assuming that the kernel \(K\) is bounded and has bounded support. For the entire collection see [Zbl 1029.00032].