Fixed design regression for time series: Asymptotic normality (Q1185836)

The authors study nonparametric regression models with dependent errors. More specifically, it is assumed that the error variables form a stationary strongly mixing process. Kernel type smoothers are studied as estimators of the regression function, including as examples the Gasser- Müller and Priestley-Chao estimators. Asymptotic normality of these estimators is established under various assumptions on the mixing rates and on the weights.