Title:ripALM: A Relative-Type Inexact Proximal Augmented Lagrangian Method with Applications to Quadratically Regularized Optimal Transport

Abstract:Inexact proximal augmented Lagrangian methods (pALMs) are particularly appealing for tackling convex constrained optimization problems because of their elegant convergence properties and strong practical performance. To solve the associated pALM subproblems, efficient methods such as Newton-type methods are essential. Consequently, the effectiveness of the inexact pALM hinges on the error criteria used to control the inexactness when solving these subproblems. However, existing inexact pALMs either rely on absolute-type error criteria (which may complicate implementation by necessitating the pre-specification of an infinite sequence of error tolerance parameters) or require an additional correction step when using relative error criteria (which can potentially slow down the convergence of the pALM). To address this deficiency, this paper proposes ripALM, a relative-type inexact pALM, which can simplify practical implementation while preserving the appealing convergence properties of the classical absolute-type inexact pALM. We emphasize that ripALM is the first relative-type inexact version of the vanilla pALM with provable convergence guarantees. Numerical experiments on quadratically regularized optimal transport (OT) problems demonstrate the competitive efficiency of the proposed method compared to existing methods. As our analysis can be extended to a more general convex constrained problem setting, including other regularized OT problems, the proposed ripALM may provide broad applicability and has the potential to serve as a basic optimization tool.