Moderate Deviations for a Test of Symmetry Based on Deconvolution Kernel Density Estimators

Suppose that Y1 , Y2 , , Yn are independent and identically distributed n observations from convolution model Y = X + ε, where X is an unobserved random variable with unknown density f X,and ε is the measurement error with a known density function. Set f n ( x )to be a nonparametric kernel density estimator of f X,and the pointwise and uniform moderate deviations of statistic sup x∈ R | f n ( x ) f n( x) |are given by Gine and Guillou’s exponential inequality.