Variable window width kernel estimates of probability densities (Q1098511)

Kernel density estimators which allow different amounts of smoothing at different locations are studied. Modifications of estimators proposed by \textit{L. Breiman}, \textit{W. Meisel} and \textit{E. Purcell} [Technometrics 19, 135-144 (1977; Zbl 0379.62023)] and \textit{I. S. Abramson} [Ann. Stat. 10, 1217-1223 (1982; Zbl 0507.62040)], which have variable window widths, are seen to have very fast rates of convergence. These rates have traditionally been obtained using a less natural higher order kernel, which has the disadvantage of allowing an estimator which takes on negative values.