English

Accelerated Single-Call Methods for Constrained Min-Max Optimization

Optimization and Control 2023-05-16 v2 Computer Science and Game Theory Machine Learning

Abstract

We study first-order methods for constrained min-max optimization. Existing methods either require two gradient calls or two projections in each iteration, which may be costly in some applications. In this paper, we first show that a variant of the Optimistic Gradient (OG) method, a single-call single-projection algorithm, has O(1T)O(\frac{1}{\sqrt{T}}) best-iterate convergence rate for inclusion problems with operators that satisfy the weak Minty variation inequality (MVI). Our second result is the first single-call single-projection algorithm -- the Accelerated Reflected Gradient (ARG) method that achieves the optimal O(1T)O(\frac{1}{T}) last-iterate convergence rate for inclusion problems that satisfy negative comonotonicity. Both the weak MVI and negative comonotonicity are well-studied assumptions and capture a rich set of non-convex non-concave min-max optimization problems. Finally, we show that the Reflected Gradient (RG) method, another single-call single-projection algorithm, has O(1T)O(\frac{1}{\sqrt{T}}) last-iterate convergence rate for constrained convex-concave min-max optimization, answering an open problem of [Heish et al, 2019]. Our convergence rates hold for standard measures such as the tangent residual and the natural residual.

Keywords

Cite

@article{arxiv.2210.03096,
  title  = {Accelerated Single-Call Methods for Constrained Min-Max Optimization},
  author = {Yang Cai and Weiqiang Zheng},
  journal= {arXiv preprint arXiv:2210.03096},
  year   = {2023}
}

Comments

Published as a conference paper at ICLR 2023

R2 v1 2026-06-28T02:57:13.617Z