A recursive quadric programming algorithm that uses new nondifferentiable penalty functions (Q1333897)

The problem considered in this paper is the equality-constrained nonlinear programming problem \[ \text{minimize } f(x)\text{ subject to } C(x)= 0,\tag{P} \] where \(f: \mathbb{R}^ n\to \mathbb{R}\), \(C: \mathbb{R}^ n\to \mathbb{R}^ m\) are twice continuously differentiable. For the solution of (P) a recursive quadratic programming algorithm is proposed which uses Han's nondifferentiable penalty functions together with a second order penalty term and Fukushima's mixed direction as the line search direction. Finally, under some particular assumptions, the method is proved to be globally convergent as well as locally convergent of second order.