Experiments on error growth associated with some linear leastsquares procedures
Author:
T. L. Jordan
Journal:
Math. Comp. 22 (1968), 579588
MSC:
Primary 65.35
MathSciNet review:
0229373
Fulltext PDF Free Access
Abstract 
References 
Similar Articles 
Additional Information
Abstract: Some numerical experiments were performed to compare the performance of procedures for solving the linear leastsquares problem based on GramSchmidt, Modified GramSchmidt, and Householder transformations, as well as the classical method of forming and solving the normal equations. In addition, similar comparisons were made of the first three procedures and a procedure based on Gaussian elimination for solving an system of equations. The results of these experiments suggest that: (1) the Modified GramSchmidt procedure is best for the leastsquares problem and that the procedure based on Householder transformations performed competitively; (2) all the methods for solving leastsquares problems suffer the effects of the condition number of , although in a different manner for the first three procedures than for the fourth; and (3) the procedure based on Gaussian elimination is the most economical and essentially, the most accurate for solving systems of linear equations. Some effects of pivoting in each of the procedures are included.
 [1]
J.
H. Wilkinson, Error analysis of direct methods of matrix
inversion, J. Assoc. Comput. Mach. 8 (1961),
281–330. MR 0176602
(31 #874)
 [2]
G.
H. Golub and J.
H. Wilkinson, Note on the iterative refinement of least squares
solution, Numer. Math. 9 (1966), 139–148. MR 0212984
(35 #3849)
 [3]
G.
Golub, Numerical methods for solving linear least squares
problems, Numer. Math. 7 (1965), 206–216. MR 0181094
(31 #5323)
 [4]
E.
E. Osborne, Smallest least squares solutions of linear
equations, J. Soc. Indust. Appl. Math. Ser. B Numer. Anal.
2 (1965), 300–307. MR 0187382
(32 #4834)
 [5]
Ȧke
Björck, Solving linear least squares problems by GramSchmidt
orthogonalization, Nordisk Tidskr. InformationsBehandling
7 (1967), 1–21. MR 0214275
(35 #5126)
 [6]
John
R. Rice, Experiments on GramSchmidt
orthogonalization, Math. Comp. 20 (1966), 325–328. MR 0192673
(33 #898), http://dx.doi.org/10.1090/S00255718196601926734
 [1]
 J. H. Wilkinson, ``Error analysis of direct methods of matrix inversion,'' J. Assoc. Comput. Mach., v. 8, 1961, pp. 281330. MR 31 #874. MR 0176602 (31:874)
 [2]
 G. Golub & J. H. Wilkinson, ``Note on the iterative refinement of least squares solution,'' Numer. Math., v. 9, 1966, pp. 139206. MR 0212984 (35:3849)
 [3]
 G. Golub, ``Numerical methods for solving linear least squares problems,'' Numer. Math., v. 7, 1965, pp. 206216. MR 31 #5323. MR 0181094 (31:5323)
 [4]
 E. E. Osbornb, ``Smallest least squares solutions of linear equations,'' SIAM J. Numer. Anal., v. 2, 1965, pp. 300307. MR 32 #4834. MR 0187382 (32:4834)
 [5]
 A. Björck, ``Solving linear least squares problems by GramSchmidt orthogonalization,'' Nordisk Tidskr. InformationsBehandling, v. 7, 1967, pp. 121. MR 0214275 (35:5126)
 [6]
 J. R. Rice, ``Experiments on GramSchmidt orthogonalization,'' Math. Comp., v. 20, 1966, pp. 325328. MR 33 #898. MR 0192673 (33:898)
Similar Articles
Retrieve articles in Mathematics of Computation
with MSC:
65.35
Retrieve articles in all journals
with MSC:
65.35
Additional Information
DOI:
http://dx.doi.org/10.1090/S0025571819680229373X
PII:
S 00255718(1968)0229373X
Article copyright:
© Copyright 1968
American Mathematical Society
