Text
New Primal-Dual Algorithms for a Class of Nonsmooth and Nonlinear Convex-Concave Minimax Problems
In this paper, we develop new primal-dual algorithms to solve a class of nonsmooth and nonlinear convex-concave minimax problems, which covers many existing and brand-new models as special cases. Our approach relies on a combination of a generalized augmented Lagrangian function, Nesterov's accelerated scheme, and adaptive parameter updating strategies. Our algorithmic framework is single-loop and unifies two important settings: general convex-concave and convex-linear cases. Under mild assumptions, our algorithms achieve convergence rates through three different criteria: primal-dual gap, primal objective residual, and dual objective residual, where is the iteration counter. Our rates are both ergodic (i.e., on a weighted averaging sequence) and nonergodic (i.e., on the last-iterate sequence). These convergence rates can be boosted up to if only one objective term is strongly convex (or, equivalently, its conjugate is -smooth). To the best of our knowledge, this is the first algorithm achieving optimal rates on the primal last-iterate sequence for convex-linear minimax problems. As a byproduct, we specify our algorithms to solve a general convex cone constrained program with both ergodic and nonergodic rate guarantees. We test our algorithms and compare them with two recent methods on two numerical examples.
Barcode | Tipe Koleksi | Nomor Panggil | Lokasi | Status | |
---|---|---|---|---|---|
art144562 | null | Artikel | Gdg9-Lt3 | Tersedia namun tidak untuk dipinjamkan - No Loan |
Tidak tersedia versi lain