We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.OC

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Optimization and Control

Title: On the Worst-Case Analysis of Cyclic Coordinate-Wise Algorithms on Smooth Convex Functions

Abstract: We propose a unifying framework for the automated computer-assisted worst-case analysis of cyclic block coordinate algorithms in the unconstrained smooth convex optimization setup. We compute exact worst-case bounds for the cyclic coordinate descent and the alternating minimization algorithms over the class of smooth convex functions, and provide sublinear upper and lower bounds on the worst-case rate for the standard class of functions with coordinate-wise Lipschitz gradients. We obtain in particular a new upper bound for cyclic coordinate descent that outperforms the best available ones by an order of magnitude. We also demonstrate the flexibility of our approach by providing new numerical bounds using simpler and more natural assumptions than those normally made for the analysis of block coordinate algorithms. Finally, we provide numerical evidence for the fact that a standard scheme that provably accelerates random coordinate descent to a $O(1/k^2)$ complexity is actually inefficient when used in a (deterministic) cyclic algorithm.
Subjects: Optimization and Control (math.OC)
Cite as: arXiv:2211.17018 [math.OC]
  (or arXiv:2211.17018v1 [math.OC] for this version)

Submission history

From: Ahmed Yassine Kamri Mr. [view email]
[v1] Wed, 30 Nov 2022 14:10:11 GMT (196kb)

Link back to: arXiv, form interface, contact.