Limit theorems for the wavelet statistics of independent random variables (Q2782709)

A sequence of \(n\) independent identically distributed random variables with a density \(f(t)\in L^2(\mathbb{R})\) is considered. The empirical estimation of \(f(t)\) in the form of wavelet-statistic is studied. There are considered three types of coefficients for these statistics (wavelet coefficients): empirical coefficients, coefficients smoothed using the procedure ``soft thresholding'', coefficients using the procedure ``hard thresholding''. The convergence of normalized estimates of \(f\) to \(N(0,1)\) is proved for three estimates of \(f\) with these three types of coefficients. The smoothing procedure improves the asymptotic properties of wavelet estimates in nonregular cases. Some other properties of estimates of \(f\) (e.g. an asymptotic unbiased estimate) and properties of coefficients (e.g. the asymptotic normality, bounds for dispersions of estimates) are also proved.