On the mean L1-error in the heteroscedastic deconvolution problem with compactly supported noises.

We study the heteroscedastic deconvolution problem when random noises have compactly supported densities. In this context, the Fourier transforms of the densities can vanish on the real line. We propose a truncated type of estimator for target density and derive the convergence rate of the mean L1-error uniformly over a class of target densities. A lower bound for the mean L1-error is also established. Some simulations will be given to illustrate the performance of the proposed estimator. [ABSTRACT FROM AUTHOR]