A new structured quasi-Newton algorithm using partial information on Hessian (Q966094): Difference between revisions

An unconstrained minimization problem on \(\mathbb{R}^n\) is solved by making use of a new modification of the structured quasi-Newton method with a vector parameter. The method uses both available gradient and function value information on Hessian of the objective function. Local and superlinear convergence of the proposed method is proved.