...
首页> 外文期刊>Chinese science bulletin >Global convergence of quasi-Newton methods for unconstrained optimization
【24h】

Global convergence of quasi-Newton methods for unconstrained optimization

机译:无约束优化的拟牛顿法的全局收敛性

获取原文
获取原文并翻译 | 示例
           

摘要

The convergence of quasi-Newton methods for unconstrained optimization has attracted much attention. Powell proved a global convergence result for the BFGS algorithm using inexact linesearch which satisfies the Wolfe conditions. Byrd, Nocedal and Yuanextended this result to the convex Broyden class of quasi-Newton methods except the DFP method. However, the global convergence of the DFP method, the first quasi-Newton method, using the same linesearch strategy, is still an open question (see ref.). Some numerical results indicate that the efficiency of the DFP method is not only much lower than that of the BFGS method, but also lower than that of other methods in the convex Broyden class (see for example refs.). This arouses suspicion: does the DFPmethod possess the same global convergence property as the BFGS method? In other words, for the DFP method, perhaps we need to add some extra conditions to get the same global convergence result. In this note, we introduce a new convergence condition, which is reasonable by our numerical experiments.
机译:拟牛顿法在无约束优化中的收敛性引起了广泛关注。 Powell使用满足Wolfe条件的不精确线搜索证明了BFGS算法的全局收敛性。 Byrd,Nocedal和Yuan将这个结果扩展到除了DFP方法之外的拟牛顿方法的凸Broyden类。但是,使用相同的线搜索策略的DFP方法(第一个拟牛顿方法)的全局收敛性仍然是一个悬而未决的问题(请参阅参考资料)。一些数值结果表明,DFP方法的效率不仅比BFGS方法低得多,而且也比凸Broyden类中的其他方法低(例如,参见参考文献)。这引起了怀疑:DFP方法是否具有与BFGS方法相同的全局收敛性?换句话说,对于DFP方法,也许我们需要添加一些额外条件才能获得相同的全局收敛结果。在本说明中,我们介绍了一个新的收敛条件,通过我们的数值实验可以证明这是合理的。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号