A class of nonmonotone stabilization trust region methods (Q1337634)

The authors present a general scheme for solving the unconstrained minimization problem `\(\min f(x)\)', \(x\in \mathbb{R}^n\), where \(f\) is a continuously differentiable function from \(\mathbb{R}^m\) to \(\mathbb{R}\). This scheme combines the nonmonotone trust region idea with the watchdog technique. Under certain conditions, one proves some convergence properties of this scheme. Extensive numerical results which are reported show that these methods are very efficient as regarding the numbers of iterations and function evaluations.