References & Citations
Mathematics > Numerical Analysis
Title: Convergence rates for an inexact ADMM applied to separable convex optimization
(Submitted on 6 Jan 2020 (this version), latest version 8 Jun 2020 (v3))
Abstract: Convergence rates are established for an inexact accelerated alternating direction method of multipliers (I-ADMM) for general separable convex optimization with a linear constraint. Both ergodic and non-ergodic iterates are analyzed. Relative to the iteration number k, the convergence rate is O(1/k) in a convex setting and O(1/k^2) in a strongly convex setting. When an error bound condition holds, the algorithm is 2-step linearly convergent. The I-ADMM is designed so that the accuracy of the inexact iteration preserves the global convergence rates of the exact iteration, leading to better performance in the test problems.
Submission history
From: William Hager [view email][v1] Mon, 6 Jan 2020 22:33:02 GMT (151kb)
[v2] Sun, 12 Jan 2020 23:21:23 GMT (151kb)
[v3] Mon, 8 Jun 2020 04:02:28 GMT (152kb)
Link back to: arXiv, form interface, contact.