Abstract:SIAM Journal on Optimization, Volume 34, Issue 2, Page 1937-1969, June 2024. In this paper, we study a class of unconstrained and constrained bilevel optimization problems in which the lower level is a possibly nonsmooth convex optimization problem, while the upper level is a possibly nonconvex optimization problem. We introduce a notion of [math]-KKT solution for them and show that an [math]-KKT solution leads to an [math]- or [math]-hypergradient–based stationary point under suitable assumptions. We also propose first-order penalty methods for finding an [math]-KKT solution of them, whose subproblems turn out to be a structured minimax problem and can be suitably solved by a first-order method recently developed by the authors. Under suitable assumptions, an operation complexity of [math] and [math], measured by their fundamental operations, is established for the proposed penalty methods for finding an [math]-KKT solution of the unconstrained and constrained bilevel optimization problems, respectively. Preliminary numerical results are presented to illustrate the performance of our proposed methods. To the best of our knowledge, this paper is the first work to demonstrate that bilevel optimization can be approximately solved as minimax optimization, and moreover, it provides the first implementable method with complexity guarantees for such sophisticated bilevel optimization.

LINEAR BILEVEL MULTIOBJECTIVE OPTIMIZATION PROBLEM: PENALTY APPROACH

A PENALTY FUNCTION METHOD FOR SOLVING NONLINEAR-LINEAR BILEVEL PROGRAMMING PROBLEM

An exact penalty method for weak linear bilevel programming problem

A Novel Penalty Approach for Nonlinear Dynamic Optimization Problems with Inequality Path Constraints

Penalty-based Methods for Simple Bilevel Optimization under Hölderian Error Bounds

A novel penalty function method for semivectorial bilevel programming problem

An objective penalty function method for nonlinear programming

On Penalty-based Bilevel Gradient Descent Method

An Evolutionary Algorithm for Nonlinear Bilevel Programming Problems Based on a New Penalty Method

A Penalty Function Algorithm with Objective Parameters for Nonlinear Mathematical Programming

First-Order Penalty Methods for Bilevel Optimization

An Alternating Linearization Method with Inexact Data for Bilevel Nonsmooth Convex Optimization

A Primal-Dual-Assisted Penalty Approach to Bilevel Optimization with Coupled Constraints

ε-exact Penalty for Non-concave Bilevel Programming Problems

Research on Bilevel Programming for Environmental Constrained Unit Commitment Optimization Problem

An Approximate Globally Convergent Method for Solving Linear Bilevel Programms

Efficient Method for Bi-level Optimization with Non-smooth Lower-Level Problem

A global solution method for semivectorial bilevel programming problem

Penalty Methods for a Class of Non-Lipschitz Optimization Problems