An application of sizing techniques to the SQP method

This paper is concerned with a sequential quadratic programming (SQP) method for nonlinear equality constrained optimization. In order to give a natural relationship between an update for the Hessian approximation and a merit function, the authors propose an SQP method that uses the structured Broyden family and the augmented Lagrangean merit function. Furthermore, they apply sizing techniques to the SQP method. Numerical results are presented to investigate how the computational performance depends on the choices of a parameter included in the structured Broyden family of a sizing factor.