Global convergence of a trust-region algorithm for inequality constrained optimization

Source abstract

This paper presents a trust-region algorithm for n-dimensional nonlinear optimization subject to m nonlinear inequality constraints. Equivalent KKT conditions are derived, being the basis for constructing the new algorithm. Global convergence of the algorithm to a first-order KKT point is established under mild conditions on the trial steps. Condition $m\leq n$ is required.