Linear Lower Bounds and Conditioning of Differentiable Games

Abstract

Recent successes of game-theoretic formulations in ML have caused a resurgence of research interest in differentiable games. Overwhelmingly, that research focuses on methods and upper bounds on their speed of convergence. In this work, we approach the question of fundamental iteration complexity by providing lower bounds to complement the linear (i.e. geometric) upper bounds observed in the literature on a wide class of problems. We cast saddle-point and min-max problems as 2-player games. We leverage tools from single-objective convex optimisation to propose new linear lower bounds for convex-concave games. Notably, we give a linear lower bound for $n$-player differentiable games, by using the spectral properties of the update operator. We then propose a new definition of the condition number arising from our lower bound analysis. Unlike past definitions, our condition number captures the fact that linear rates are possible in games, even in the absence of strong convexity or strong concavity in the variables.

arXiv commentICML 2020 final version

Cited by in corpus (9)

• Smooth Monotone Stochastic Variational Inequalities and Saddle Point Problems: A Survey
• The Complexity of Nonconvex-Strongly-Concave Minimax Optimization
• Complexity Lower Bounds for Nonconvex-Strongly-Concave Min-Max Optimization
• Solving Structured Hierarchical Games Using Differential Backward Induction
• Near-optimal Local Convergence of Alternating Gradient Descent-Ascent for Minimax Optimization
• A Tight and Unified Analysis of Gradient-Based Methods for a Whole Spectrum of Games
• Near Optimal Stochastic Algorithms for Finite-Sum Unbalanced Convex-Concave Minimax Optimization
• Minimax Optimization: The Case of Convex-Submodular
• Complex Momentum for Optimization in Games