1991, 14(1):33-41.
Abstract:
A new trust region algorithm is proposed for solving unconstrained optimization problems. According to the quadratic approximate model of the original optimization problem,the trust region algorithm uses directions,a convex combination of the quasi-newton direction and the steepest descent direction. This algorithm with new strategy is analyzed and the global and local quadratic convergent theorems are proved. At last,the implementation and computational results of the algorithm are demonstrated.