| 摘要: |
| 基于锥模型的拟牛顿法已被许多研究者讨论过，并且D．C．Sorensen文（TheQ-superlinearconvergenceofacolllnearscalingalgorithmforunconstrainedoptimization．SIAMJNumerAnal，1980，17（1）：84～114）证明了该算法模型是超线性收敛的。本文中针对1维优化问题讨论了该算法模型的收敛阶，结果表明它是小Q-2阶收敛的，并且从极小点X的左右两边交错收敛到X。 |
| 关键词: 1维优化,锥模型方法,收敛阶 |
| DOI： |
|
| 基金项目:国家自然科学基金 |
|
| Convergence Rate of Conic Methods for One-dimensional Unconstrained Optimization |
|
|
|
| Abstract: |
| Conic methods for unconstrained optimization have been discussed by manypeople, and its superlinear convergence rate has been proved by D. C. Sorensen. In thispaper, it is proved that the convergence rate order of conic method for one-dimensionaloptimization is Q -2 and converges to x' from two sides alternately. |
| Key words: one-dimensional optimization,conic method,convergence rate, |
