Revisiting the Geometrically Decaying Step Size: Linear Convergence for Smooth or Non-Smooth Functions

Fuente: arXiv
Saved in:
Bibliographic Details
Main Author: Kim, Jihun
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917120605945856
author Kim, Jihun
author_facet Kim, Jihun
contents We revisit the geometrically decaying step size given a positive inverse condition number, under which a locally Lipschitz function shows linear convergence. The positivity does not require the function to satisfy convexity, weak convexity, quasar convexity, or sharpness, but instead amounts to a property strictly weaker than the assumptions used in existing works (e.g., weak convexity + sharpness). We propose a clean and simple subgradient descent algorithm that requires minimal knowledge of problem constants, applicable to either smooth or non-smooth functions.
format Preprint
id arxiv_https___arxiv_org_abs_2508_13569
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Revisiting the Geometrically Decaying Step Size: Linear Convergence for Smooth or Non-Smooth Functions
Kim, Jihun
Optimization and Control
49J52, 90C26, 90C30, 90C56
We revisit the geometrically decaying step size given a positive inverse condition number, under which a locally Lipschitz function shows linear convergence. The positivity does not require the function to satisfy convexity, weak convexity, quasar convexity, or sharpness, but instead amounts to a property strictly weaker than the assumptions used in existing works (e.g., weak convexity + sharpness). We propose a clean and simple subgradient descent algorithm that requires minimal knowledge of problem constants, applicable to either smooth or non-smooth functions.
title Revisiting the Geometrically Decaying Step Size: Linear Convergence for Smooth or Non-Smooth Functions
topic Optimization and Control
49J52, 90C26, 90C30, 90C56
url https://arxiv.org/abs/2508.13569