A feasible descent SQP algorithm for general constrained optimization without strict complemen\-tar\-ity (Q556270)

Smooth optimization problems with equality and inequality constraints are considered. First, the problem is transformed into one where only inequality constraints and a parameter in the objective function are involved as suggested by \textit{D. Q. Mayne} and \textit{E. Polak} [Math. Program. 11, 67--80 (1976; Zbl 0351.90067)]. Then the simplified problem is solved by a successive quadratic programming (SQP) method, which is based on \(\varepsilon\)-active constraints. The search directions that result from the quadratic subproblem are adjusted twice, once to get a feasible direction once to avoid the Maratos effect. Global and superlinear convergence is proven when strict complementarity is not used.