“…While the univariate Thorin class benefits from a large description of its content and properties, e.g., in [25,36], or in [42,27,7,8], the statistical literature and especially regarding the problem of the estimation of such parametric models is a lot sparser. One potential reason is that the deconvolution of an empirical distribution into several Gamma distributions is a hard inverse problem, and the deconvolution literature already has troubles with removing a small convolutional noise from a strong signal.…”