Trust-region and other regularisations of linear least-squares problems

被引:23
|
作者
Cartis, C. [1 ]
Gould, N. I. M. [2 ]
Toint, P. L. [3 ]
机构
[1] Univ Edinburgh, Sch Math, Edinburgh EH9 3JZ, Midlothian, Scotland
[2] Rutherford Appleton Lab, Computat Sci & Engn Dept, Chilton OX11 0QX, Oxon, England
[3] Univ Namur, Fac Univ Notre Dame Paix, Dept Math, B-5000 Namur, Belgium
基金
英国工程与自然科学研究理事会;
关键词
Linear least-squares; Regularisation; Trust-region; Secular equation; ALGORITHM; REGULARIZATION; EQUATIONS; LSQR;
D O I
10.1007/s10543-008-0206-8
中图分类号
TP31 [计算机软件];
学科分类号
081202 ; 0835 ;
摘要
We consider methods for regularising the least-squares solution of the linear system Ax = b. In particular, we propose iterative methods for solving large problems in which a trust-region bound ||x|| <= Delta is imposed on the size of the solution, and in which the least value of linear combinations of ||Ax - b||(q)(2) and a regularisation term ||x||(2)(p) for various p and q = 1, 2 is sought. In each case, one or more "secular" equations are derived, and fast Newton-like solution procedures are suggested. The resulting algorithms are available as part of the GALAHAD optimization library.
引用
收藏
页码:21 / 53
页数:33
相关论文
共 50 条