首页> 外文会议>The Fifth China-Japan Joint Seminar on Numerical Mathematics Aug 21-25, 2002 Shanghai, China >Projection methods for the least squares problems with quadratic equality constraints
【24h】

Projection methods for the least squares problems with quadratic equality constraints

机译:具有二次等式约束的最小二乘问题的投影方法

获取原文
获取原文并翻译 | 示例

摘要

We consider the quadratic equality constrained least squares problem of minimizing ‖Ax - b‖_2 subject to the constraint ‖x‖_2 = 1, without the assumption ‖A~+b‖_2 > 1 which is commonly imposed in literatures. A perturbation analysis is first given to show the sensitivity of the problem. We propose a projection method and relative correction approaches for solving this problem. We also give a detailed convergence analysis of our proposed algorithms to illustrate the convergence behavior. Numerical experiments show that our projection method with certain correction techniques converges at a high percentage over 90% among all the test numerical experiments while Newton-like methods commonly in use almost always fails.
机译:我们考虑在约束条件“ x” _2 = 1的情况下,使“ Ax-b” _2最小化的二次等式约束最小二乘问题,而没有假设文献中通常采用“ A〜+ b” _2> 1的假设。首先进行扰动分析以显示问题的敏感性。我们提出了一种投影方法和相对校正方法来解决这个问题。我们还对提出的算法进行了详细的收敛分析,以说明收敛行为。数值实验表明,在所有测试数值实验中,采用某些校正技术的投影方法在90%以上会以较高的比例收敛,而通常使用的类似牛顿法几乎总是会失败。

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