作者: B. Jin , T. Takeuchi
DOI: 10.1080/02331934.2015.1101598
关键词:
摘要: In this paper, we revisit the augmented Lagrangian method for a class of nonsmooth convex optimization. We present Lagrange optimality system associated with problems, and establish its connections standard condition saddle point Lagrangian, which provides powerful tool developing numerical algorithms: derive Lagrange–Newton algorithm optimization, nonsingularity Newton local convergence algorithm.