| 摘要: |
| 给出了非线性最小二乘问题的结构p步牛顿法算法，分析了该算法的效率，结果表明，对零残差问题新算法具有q-2阶收敛速率，与牛顿法具有相同的收敛速率，由于新算法只需计算近似海赛矩阵，所以，其效应比牛顿法高，对于非零残差问题算法具有p步p 1阶收敛速率，其效率至少与牛顿法相同。 |
| 关键词: 最小二乘问题 P步牛顿法 效率 |
| DOI： |
| 修订日期：1999-11-30 |
| 基金项目:国家自然科学基金资助项目 |
|
| Structured p-step Newton Algorithm for Nonlinear Least Square Problems |
|
|
|
| Abstract: |
| A structured p step Newton algorithm for nonlinear least square problems is developed. The efficiency of the algorithm is analysed. For zero residual problem its convergence rate is q 2 order, which is the same with that of Newton algorithm. Since the Hessian matrix is calculated appropriately, the efficiency of this algorithm is higher than that of Newton method; For non zero residual problem its convergence rate is p step p 1 order, the efficiency of this algorithm is at least as high as that of Newton method. |
| Key words: Least square problem,p-step Newton algorithm,efficiency |
