Abstract: In this talk, we introduce a gradient method based on a quadratic approximation model for unconstrained optimization problems, and prove the convergence of  the method.