Obtaining the kinetic function of depolymerization from evolving molecular weight distribution data–an inverse problem