No label defined (Q4791815)

The authors deal with the asymptotic behaviour of the kernel estimates \[ f_n(x)=(|U_n|h_n)^{-1}\sum_{j\in U_n}K((x-X_j)h_n^{-1}),\quad x\in{\mathbb R},\;n\in{\mathbb N}, \] where \(K\) is a density of a given distribution (kernel), \(\{h_n\},\;n\geq 1\), is a sequence of positive numbers approaching zero as \(n\to\infty\), and \(|U_n|\) is the number of elements of the finite set \(U_n\). They propose estimates for the rate of convergence \[ \Delta_n(x):=\sup_{z\in\mathbb R}|P(A_n(x)<z)-\Phi(z)| \] of the normalized kernel estimates \[ A_n(x)\colon=(f_n(x)-Ef_n(x))/\sqrt{\text{Var}(f_n(x))} \] to the standard normal distribution for kernel density estimates of the real valued quasi-associated random field \(X=\{X_j,\;j\in\mathbb Z^d\}\).