Second-Order Optimality Conditions for Sparse Optimization via Fréchet Second-Order Subdifferential
摘要
In this paper, we investigate the second-order optimality conditions of sparsity-constrained optimization problems under the Fréchet second-order subdifferential of its Lagrangian function assumption instead of the twice continuous differentiability of the objective and the constraint functions. The second-order necessary optimality conditions of the sparse optimization problem are established under suitable Robinson constraint qualifications. Further, the second-order sufficient optimality conditions of the sparse optimization problem with equality constraints are also derived under mild conditions. The results presented in this paper extend the second-order optimality conditions [SIAM J. Optim. 2, 766-784 (2023)] to sparse optimization.