In this paper, a nonmonotone line search strategy is presented for minimization of the locally Lipschitz continuous function. First, the Armijo condition is generalized along a descent direction at the current point. Then, a step length is selected along a descent direction satisfying the generalized Armijo condition. We show that there exists at least one step length satisfying the generalized Armijo condition. Next, the nonmonotone line search algorithm is proposed and its global convergence is proved. Finally, the proposed algorithm is implemented in the MATLAB environment and compared with some methods in the subject literature. It can be seen that the proposed method not only computes the global optimum also reduces the number of function evaluations than the monotone line search method.
