Minimax Theory and its Applications 08 (2023), No. 2, 333--380
Copyright Heldermann Verlag 2023
Last-Iterate Convergence of Saddle-Point Optimizers via High-Resolution Differential Equations
Tatjana Chavdarova
EECS, University of California, Berkeley, U.S.A.
tatjana.chavdarova@berkeley.edu
Michael I. Jordan
EECS, University of California, Berkeley, U.S.A.
jordan@cs.berkeley.edu
Manolis Zampetakis
EECS, University of California, Berkeley, U.S.A.
mzampet@berkeley.edu
Several widely-used first-order saddle-point optimization methods yield an identical continuous-time ordinary differential equation (ODE) that is identical to that of the Gradient Descent Ascent (GDA) method when derived naively. However, the convergence properties of these methods are qualitatively different, even on simple bilinear games. Thus the ODE perspective, which has proved powerful in analyzing single-objective optimization methods, has not played a similar role in saddle-point optimization.
We adopt a framework studied in fluid dynamics -- known as High-Resolution Differential Equations (HRDEs) -- to design differential equation models for several saddle-point optimization methods. Critically, these HRDEs are distinct for various saddle-point optimization methods. Moreover, in bilinear games, the convergence properties of the HRDEs match the qualitative features of the corresponding discrete methods. Additionally, we show that the HRDE of Optimistic Gradient Descent Ascent (OGDA) exhibits last-iterate convergence for general monotone variational inequalities. Finally, we provide rates of convergence for the best-iterate convergence of the OGDA method, relying solely on the first-order smoothness of the monotone operator.
Keywords: Variational inequality, convergence, high resolution differential equations, saddle-point optimizers, continuous time methods.
MSC: 34H05, 49M15, 90C30.
[ Fulltext-pdf (612 KB)] for subscribers only.