A modified class of nonmonotone Armijo-type line search method for unconstrained optimization

In this paper, we propose a new family of line search method for unconstrained optimization problems in that we incorporate a modified Armijo line search rule with nonmonotone strategy. Right hand side of new modified Armijo rule is greater than standard Armijo rule, so possibly we can get a larger step size and decrease the number of iterations and function evaluations. Moreover, we prove the global convergence of the new algorithm, and also we analyze the convergence rate of this approach under some suitable conditions. Preliminary numerical experiments establish that the new approach is robustness and efficient for unconstrained optimization problems.

Keywords:

Unconstrained optimization , Armijo-type line search , Nonmonotone strategy , Globalconvergence