A globally convergent trust-region SQP method without a penalty function for nonlinearly constrained optimization

In this paper, we propose a new trust-region SQP method, which uses no penalty function, for solving nonlinearly constrained optimization problem. Our method consists of alternate two phases. Specifically, we alternately proceed the feasibility restoration phase and the objective function minimization phase. The global convergence property of the proposed method is shown.

Citation

Cooperative Research Report 168 "OPTIMIZATION: Modeling and Algorithms 17", The Institute of Statistical Mathematics, Tokyo, Japan, February 2004

Article

Download

View A globally convergent trust-region SQP method without a penalty function for nonlinearly constrained optimization