| 不等式约束优化拟正则条件下的超线性收敛FSSLE算法 |
| |
| 引用本文： | 张新华,李学全.不等式约束优化拟正则条件下的超线性收敛FSSLE算法[J].经济数学,2009,26(3):85-94. |
| |
| 作者姓名： | 张新华 李学全 |
| |
| 作者单位： | 1. 南京农业大学工学院,江苏,南京,210031
2. 湖南第一师范学院,湖南,长沙,410083 |
| |
| 基金项目： | 江苏省农机局科研基金，湖南省教育厅科学研究项目 |
| |
| 摘 要： | 对不等式约束优化问题。提出一个可行序列线性方程组（FSSLE）算法。该算法每次迭代只需求解两个具有相同系数矩阵的线性方程组，因而计算量较小。在一定条件下，算法具有全局收敛性。在没有严格互补条件、比强二阶充分条件弱的拟正则条件下，证明了算法具有超线性收敛性并用数值试验表明其有效性。
|
| 关 键 词： | 非线性规划 FSSLE 全局收敛 超线性收敛 |
A SUPERLINEARLY CONVERGENT FSSLE ALGORITHM FOR INEQUALITY CONSTRAINED OPTIMIZATION WITH QUASI-REGULARITY |
| |
| Institution: | ZHANG Xin-hua, LI Xue-quan( 1. College of Engineering, Nanjing Agricultural University, Nanjing 210031 ;2. Hunan First Normal University, Changsha 410083) |
| |
| Abstract: | A feasible sequential system of linear equations （FSSLE） algorithm was proposed for solving inequality constrained optimization problems. It is observed that the algorithm is merely necessary to solve two systems of linear equations with the same coefficient matrix. Under some suitable conditions, we show that this algorithm is globally convergence. Without assuming the strict complementarity, the superlinear conver gence rate is obtained under the quasi-regularity, which is weaker than the strong second-order sufficiency condition. Numerical experiments show that the method is effective. |
| |
| Keywords: | FSSLE |
| 本文献已被 维普 万方数据 等数据库收录！ |
|
