Central limit theorems for \(L_ p\) distances of kernel estimators of densities under random censorship (Q1184207)

\(X^ 0_ 1,X^ 0_ 2,\dots\) is a sequence of i.i.d. nonnegative random variables, with common cdf \(F\), density function \(f\). \(Y_ 1,Y_ 2,\dots\) is a sequence of i.i.d. continuous random variables, independent of \(X^ 0_ 1,X^ 0_ 2,\dots\), and censoring \(X^ 0_ 1,X^ 0_ 2,\dots\) on the right, so that we observe the pairs \((X_ 1,d_ 1),\dots,(X_ n,d_ n)\), where \(X_ j=\min(X^ 0_ j,Y_ j)\) and \(d_ j\) is the indicator of the event \((X_ j=X^ 0_ j)\). \(\hat F_ n\) denotes the Kaplan-Meier product-limit estimator of \(F\). The density \(f\) is estimated by a sequence of kernel estimators \(f_ n(t)=[\int_{- \infty}^ \infty K((t-x)/h(n))d\hat F_ n(x)]/h(n)\), where \(h(n)\) is a sequence of numbers and \(K\) is a kernel density function. Central limit theorems are proved for \(\int_ 0^ T| f_ n(t)-f(t)|^ p du(t)\) where \(1\leq p<\infty\), \(0<T\leq\infty\), and \(u\) is a measure on the Borel sets of the real line. The results are applied to tests of goodness of fit.