Global convergence of the non-quasi-Newton method for unconstrained optimization problems (Q854562)

The authors investigate the properties of non-quasi-Newton algorithmic methods in unconstrained problems. While quasi-Newton methods are known to be efficient, the same is not always true for non-quasi-Newton methods. The authors begin by presenting in detail a new non-quasi-Newton algorithm (in Section 2) which is studied in detail in Section 3. It is proven that the algorithm is globally convergent and, under some conditions, superlinear convergent. The paper concludes with a series of numerical results and a list of relevant references.