We estimate the distribution of a real-valued random variable from contaminated observations. The additive error is supposed to be normally distributed, but with an unknown variance. The distribution is identifiable from the observations if we restrict the class of considered distributions by a...

The purpose of this work is to provide upper bounds on the stop-loss and total variation distances between random sums. The main theoretical argument consists in defining discrete analogs of the classical ideal metrics considered by Rachev and Rüschendorf (Adv. Appl. Probab. 22 (1990) 350). An...