Quadratically constraint quadratical algorithm model for nonlinear minimax problems (Q2378762)

The quadratically approximate algorithm framework for solving general constrained minimax problems, presented in this paper, is a very general form for minimax problems and general nonlinear programs, and many sequential quadratic programming (SQP) methods, sequential quadratically constrained quadratic programming (SQCQP) methods and norm-relaxed methods can be regarded as special cases of the algorithm framework. So, the results presented in this paper could be used to analyze the convergence rate of SQP methods, SQCQP methods and norm-relaxed methods for minimax problems and general nonlinear programs. The global convergence of the algorithm framework is obtained under the Mangasarian-Fromovitz constraint qualification (MFCQ), and the conditions for superlinear convergence of the algorithm framework are presented under the MFCQ, the constant rank constraint qualification as well as the strong second-order sufficiency conditions (SSOSC). Also quadratic convergence rate is obtained under the MFCQ and SSOSC.