Global saddle points of nonlinear augmented Lagrangian functions

We notice that the results for the existence of global (local) saddle points of augmented Lagrangian functions in the literature were only sufficient conditions of some special types of augmented Lagrangian. In this paper, we introduce a general class of nonlinear augmented Lagrangian functions for constrained optimization problem. In two different cases, we present sufficient and necessary conditions for the existence of global saddle points. Moreover, as corollaries of the two results above, we not only obtain sufficient and necessary conditions for the existence of global saddle points of some special types of augmented Lagrangian functions mentioned in the literature, but also give some weaker sufficient conditions than the ones in the literature. Compared with our recent work (Wang et al. in Math Oper Res 38:740–760, 2013), the nonlinear augmented Lagrangian functions in this paper are more general and the results in this paper are original. We show that some examples (such as improved barrier augmented Lagrangian) satisfy the assumptions of this paper, but not available in Wang et al. (2013).

[1]  M. Hestenes.  Multiplier and gradient methods , 1969 .

[2]  R. Rockafellar.  Augmented Lagrange Multiplier Functions and Duality in Nonconvex Programming , 1974 .

[3]  K. I. M. McKinnon,et al.  On Saddle Points of Augmented Lagrangians for Constrained Nonconvex Optimization , 2005, SIAM J. Optim..

[4]  Dimitri P. Bertsekas,et al.  Constrained Optimization and Lagrange Multiplier Methods , 1982 .

[5]  X. Q. Yang,et al.  Some Results about Duality and Exact Penalization , 2004, J. Glob. Optim..

[6]  X. Q. Yang,et al.  Nonlinear Augmented Lagrangian and Duality Theory , 2013, Math. Oper. Res..

[7]  R. Tyrrell Rockafellar,et al.  Variational Analysis , 1998, Grundlehren der mathematischen Wissenschaften.

[8]  H. Lee,et al.  On saddle points of a class of augmented lagrangian functions , 2007 .

[9]  Petr Mandl.  Lagrange multipliers and optimization of non-terminating Markov chains , 1969, Kybernetika.

[10]  Duan Li,et al.  Unified theory of augmented Lagrangian methods for constrained global optimization , 2009, J. Glob. Optim..

[11]  Alfredo N. Iusem,et al.  Duality and Exact Penalization for General Augmented Lagrangians , 2010, J. Optim. Theory Appl..

[12]  Tao Li,et al.  Local saddle point and a class of convexification methods for nonconvex optimization problems , 2007, J. Glob. Optim..

[13]  H. X. Wu,et al.  Saddle points of general augmented Lagrangians for constrained nonconvex optimization , 2012, J. Glob. Optim..

[14]  M. J. D. Powell,et al.  A method for nonlinear constraints in minimization problems , 1969 .

[15]  Xiaoqi Yang,et al.  A Unified Augmented Lagrangian Approach to Duality and Exact Penalization , 2003, Math. Oper. Res..

[16]  Jinchuan Zhou,et al.  Saddle Points Theory of Two Classes of Augmented Lagrangians and Its Applications to Generalized Semi-infinite Programming , 2009 .

[17]  R. Tyrrell Rockafellar,et al.  Lagrange Multipliers and Optimality , 1993, SIAM Rev..

[18]  Constantin Zalinescu,et al.  A Nonlinear Extension of Hoffman's Error Bounds for Linear Inequalities , 2003, Math. Oper. Res..

[19]  Naihua Xiu,et al.  Saddle point and exact penalty representation for generalized proximal Lagrangians , 2012, J. Glob. Optim..