A unified analysis on the speedup of accelerated gradient methods: An inertial dynamics approach

Zepeng Wang, Juan Peypouquet · 2025

Read A unified analysis on the speedup of accelerated gradient methods: An inertial dynamics approach by Zepeng Wang, Juan Peypouquet (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-16953.txt.

Description

Zepeng Wang and Juan Peypouquet explain the speed of fast gradient methods in a 2025 paper. Gradient methods minimize a function by stepping downhill, and accelerated versions add momentum to converge far faster. The best recent variants run twice as fast as the classic method, but their proofs rest on heavy machinery that hides the reason.

The authors trace the speedup to fine links between coefficients in a damped inertial system. A standard discretization of that system yields intuitive variants of the classic method with the faster rate. Their proof reaches beyond strongly convex cases to functions with quadratic growth or a related slope condition. The work gives designers a clearer physical picture of why momentum helps.

Catalog record

CREATORZepeng Wang, Juan Peypouquet
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE94 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2502-16953

Browse the full library · About the Tower of Babel