Modifying feasible SQP method for inequality constrained optimization

Abstract

This paper is concerned with an improved feasible sequential quadratic programming (FSQP) method which solves an inequality constrained nonlinear optimization problem. As compared with the existing SQP methods, at each iteration of our method, the base direction is only necessary to solve a equality constrained quadratic programming, the feasible direction and the high-order revised direction which avoids Maratos effect are obtained by explicit formulas. Furthermore, the global and superlinear convergence are proved under some suitable conditions. © 2012 Springer-Verlag.