On the approximate augmented Lagrangian for nonlinear symmetric cone programming

Yong-Jin Liu,Li-Wei Zhang

DOI: https://doi.org/10.1016/j.na.2006.12.016

2008-01-01

Abstract:This paper studies the approximate augmented Lagrangian for nonlinear symmetric cone programming. The analysis is based on some results under the framework of Euclidean Jordan algebras. We formulate the approximate Lagrangian dual problem and study conditions for approximate strong duality results and an approximate exact penalty representation. We also show, under Robinson’s constraint qualification, that the sequence of stationary points of the approximate augmented Lagrangian problems converges to a stationary point of the original nonlinear symmetric cone programming.