accelerated gradient methods play a central role in optimization, achieving
optimal rates in many settings. While many generalizations and extensions of
Nesterov's original acceleration method have been proposed, it is not yet clear
what is the natural scope of the acceleration concept