English

A Dynamical Systems Perspective on Nesterov Acceleration

Optimization and Control 2019-05-21 v1 Machine Learning Systems and Control Machine Learning

Abstract

We present a dynamical system framework for understanding Nesterov's accelerated gradient method. In contrast to earlier work, our derivation does not rely on a vanishing step size argument. We show that Nesterov acceleration arises from discretizing an ordinary differential equation with a semi-implicit Euler integration scheme. We analyze both the underlying differential equation as well as the discretization to obtain insights into the phenomenon of acceleration. The analysis suggests that a curvature-dependent damping term lies at the heart of the phenomenon. We further establish connections between the discretized and the continuous-time dynamics.

Keywords

Cite

@article{arxiv.1905.07436,
  title  = {A Dynamical Systems Perspective on Nesterov Acceleration},
  author = {Michael Muehlebach and Michael I. Jordan},
  journal= {arXiv preprint arXiv:1905.07436},
  year   = {2019}
}

Comments

11 pages, 4 figures, to appear in the Proceedings of the 36th International Conference on Machine Learning

R2 v1 2026-06-23T09:11:11.003Z