Abstract: | We introduce a new method for solving an unconstrained optimization problem. The method is based on the solution of a variational inequality problem and may be considered a modified trust region algorithm. We prove the convergence of the method and present some numerical results. |