Control Theory-Inspired Acceleration of the Gradient-Descent Method: Centralized and Distributed