Abstract
For least squares problems of minimizing ∥b −Ax∥2 whereA is a large sparsem ×n (m ≥n) matrix, the common method is to apply the conjugate gradient method to the normal equationA T Ax =A T b. However, the condition number ofA T A is square of that ofA, and convergence becomes problematic for severely ill-conditioned problems even with preconditioning. In this paper, we propose two methods for applying the GMRES method to the least squares problem by using ann ×m matrixB. We give the necessary and sufficient condition thatB should satisfy in order that the proposed methods give a least squares solution. Then, for implementations forB, we propose an incomplete QR decomposition IMGS(l). Numerical experiments showed that the simplest casel = 0 gives the best results, and converges faster than previous methods for severely ill-conditioned problems. The preconditioner IMGS(0) is equivalent to the caseB = (diag(A T A))−1 A T, so (diag(A T A))−1 A T was the best preconditioner among IMGS(l) and Jennings’ IMGS(τ). On the other hand, CG-IMGS(0) was the fastest for well-conditioned problems.
Similar content being viewed by others
References
M. Benzi and M. Tuma, A robust preconditioner with low memory requirements for large sparse least squares problems. SIAM J. Sci. Comput.,25 (2003), 499–512.
A. Björck, Numerical Methods for Least Squares Problems. SIAM, Philadelphia, 1996.
P.N. Brown and H.F. Walker, GMRES on (nearly) singular systems. SIAM J. Matrix Anal. Appl.,18 (1997), 37–51.
D. Calvetti, B. Lewis, and L. Reichel, GMRES-type methods for inconsistent systems. Linear Algebra Appl.,316 (2000), 157–169.
M.R. Hestenes and E. Stiefel, Methods of conjugate gradients for solving linear systems. J. Res. Natl. Bur. Stand.,49 (1952), 409–436.
A. Jennings and M.A. Ajiz, Incomplete methods for solving AT Ax =b. SIAM J. Sci. Stat. Comput.,5 (1984), 978–987.
National Institute of Standards, Matrix Market: test matrices database, available on line at http://math.nist.gov/MatrixMarket/data/, Gaithersburg, MD.
Y. Saad, Preconditioning techniques for nonsymmetric and indefinite linear systems. J. Comput. Appl. Math.,24 (1988), 89–105.
Y. Saad and M.H. Schultz, GMRES: A generalized minimal residual algorithm for solving nonsymmetric linear systems. SIAM J. Sci. Stat. Comput.,7 (1986), 856–869.
S.-L. Zhang, Generalization of the Conjugate Residual Method. Ph.D. Thesis, University of Tsukuba, Japan, 1989 (in Japanese).
S.-L. Zhang and Y. Oyanagi, A necessary and sufficient convergence condition of orthomin(κ) methods for least squares problem with weight. Ann. Inst. Statist. Math.,42 (1990), 805–811.
S.-L. Zhang and Y. Oyanagi, Orthomin(κ) method for linear least squares problem. J. Inf. Proc.,14 (1991), 121–125.
Author information
Authors and Affiliations
Corresponding author
Additional information
The research of this author was supported by the Grant-in-Aid for Scientific Research of the Ministry of Education, Culture Sports, Science and Technology, Japan.
About this article
Cite this article
Ito, T., Hayami, K. Preconditioned GMRES methods for least squares problems. Japan J. Indust. Appl. Math. 25, 185 (2008). https://doi.org/10.1007/BF03167519
Received:
Revised:
DOI: https://doi.org/10.1007/BF03167519