Cyclic Barzilai-Borwein Gradient Method with Retards Step-size
Gradient method is a common method for solving large scale unconstrained optimization problems.In this paper,the step size of solving quadratic function minimization problem is extended to solving general unconstrained optimization problem.By using the idea of one-step retard and cyclic gradient method,the cyclic BB gradient method is proposed,and combined with the Zhang-Hager non-monotonic line search technique,A cyclic BB gradient algorithm-CBBGM algorithm is presented for solving gen-eral unconstrained optimization problems.Under suitable assumptions,the CBBGM algorithm has global convergence and linear con-vergence rate when the objective function is strongly convex.Numerical experiments show that the proposed method is more computa-tionally efficient than the existing methods.
Barzilai-Borwein gradient methodunconstrained optimization problemZhang-Hager non-monotonic line searchglobal convergence
万方数据
维普
