A GEOMETRIC PROPERTY OF THE LEAST-SQUARES SOLUTION OF LINEAR-EQUATIONS

被引：31

作者：

BENTAL, A ^{[1
]}

TEBOULLE, M ^{[1
]}

机构：

[1] UNIV MARYLAND,DEPT MATH & STAT,CATONSVILLE,MD 21228

来源：

LINEAR ALGEBRA AND ITS APPLICATIONS | 1990年 / 139卷

关键词：

D O I：

10.1016/0024-3795(90)90395-S

中图分类号：

O29 [应用数学];

学科分类号：

070104 ;

摘要：

We derive an explicit determinantal formula for the least squares solution of an overdetermined system of linear equations. From this formula it follows that the least squares solution lies in the convex hull of the solutions to the square subsystems of the original system. We extend this result and prove that this geometric property holds for a more general class of problems which includes the weighted least squares and lp-norm minimization problems. © 1990.

引用

页码：165 / 170

页数：6

共 7 条

[1]

BENISRAEL A, 1974, GENERALIZED INVERSES

[2]

Gantmacher, 1959, THEORY MATRICES, V2

[3] ON PROJECTED NEWTON BARRIER METHODS FOR LINEAR-PROGRAMMING AND AN EQUIVALENCE TO KARMARKAR PROJECTIVE METHOD [J].

GILL, PE ;

MURRAY, W ;

SAUNDERS, MA ;

TOMLIN, JA ;

WRIGHT, MH .

MATHEMATICAL PROGRAMMING, 1986, 36 (02) :183-209

[4]

Goldman AJ, 1956, ANN MATH STUD, V38, P41

[5] A NEW POLYNOMIAL-TIME ALGORITHM FOR LINEAR-PROGRAMMING [J].

KARMARKAR, N .

COMBINATORICA, 1984, 4 (04) :373-395

[6]

LINNIK YV, 1961, METHOD LEAST SQUARES

[7]

Ortega J.M., 1970, OCLC1154227410, Patent No. 1154227410

← 1 →