Global convergence of nonmonotone descent methods for unconstrained optimization problems (Q697549)

This paper deals with global convergence for unconstrained optimization algorithms that utilize a nonmonotone line search procedure. The proposed technique allows the user to specify a flexible facing function and includes Armijo, Goldstein and Wolfe rules as special cases. The global convergence (towards stationary points) of the nonmonotone descent method for unconstrained optimization is proved.