Numerical behavior of a stabilized SQP method for degenerate NLP problems

In this paper we discuss the application of the stabilized SQP method with constraint identification (sSQPa) recently proposed by S. J. Wright [12] for nonlinear programming problems at which strict complementarity and/or linear independence of the gradients of the active constraints may fail to hold at the solution. We have collected a number of degenerate problems from different sources. Our numerical experiments have shown that the sSQPa is efficient and robust even without the incorporation of a classical globalization technique. One of our goals is therefore to handle NLPs that arise as subproblems in global optimization where degeneracy and infeasibility are important issues. We also discuss and present our work along this direction.