摘要 : A local algorithm is proposed for unconstrained optimization problem. Compared with the traditional Newton method with Choleski factorization, this algorithm has the same quadratic convergence. But its computation cost per iterati... 展开 A local algorithm is proposed for unconstrained optimization problem. Compared with the traditional Newton method with Choleski factorization, this algorithm has the same quadratic convergence. But its computation cost per iteration in average is less when the dimension n≥55. The saving is estimated in the theoretical framework. 收起