A feasible SQP method for nonlinear programming (Q2266975)

The authors consider general nonlinear programming problems with both equality and inequality constraints. Existing Sequential Quadratic Methods (SQM) are surveyed. A serious drawback of the SQM consists in the necessity to solve a quadratic subproblem with inequality constraints. This fact motivated the authors to propose an algorithm, which overcomes this drawback. The proposed algorithm requires only solving linear systems or equivalently equality constrained quadratic subproblems on each iteration. Global and superlinear convergence under appropriate assumptions is proved.