Abstract
For the large-scale linear discrete ill-posed problem \(\min \limits \|Ax-b\|\) or Ax = b with b contaminated by Gaussian white noise, the following Krylov solvers are commonly used: LSQR, and its mathematically equivalent CGLS (i.e., the Conjugate Gradient (CG) method applied to ATAx = ATb), CGME (i.e., the CG method applied to \(\min \limits \|AA^{T}y-b\|\) or AATy = b with x = ATy), and LSMR (i.e., the minimal residual (MINRES) method applied to ATAx = ATb). These methods have intrinsic regularizing effects, where the number k of iterations plays the role of the regularization parameter. In this paper, we analyze the regularizing effects of CGME and LSMR and establish a number of results including the filtered SVD expansion of CGME iterates, which prove that the 2-norm filtering best possible regularized solutions by CGME and LSMR are less accurate than and at least as accurate as those by LSQR, respectively. We also prove that the semi-convergence of CGME and LSMR always occurs no later and sooner than that of LSQR, respectively. As a byproduct, using the analysis approach for CGME, we improve a fundamental result on the accuracy of the truncated rank k approximate SVD of A generated by randomized algorithms, and reveal how the truncation step damages the accuracy. Numerical experiments justify our results on CGME and LSMR.
Similar content being viewed by others
References
Aster, R.C., Borchers, B., Thurber, C.H.: Parameter Estimation and Inverse Problems, 2nd edn. Elsevier, New York (2013)
Berisha, S., Nagy, J.G.: Restore tools: Iterative methods for image restoration. Available from http://www.mathcs.emory.edu/~nagy/RestoreTools (2012)
Björck, Å: Numerical Methods for Least Squares Problems. SIAM, Philadelphia (1996)
Björck, Å: Numerical Methods in Matrix Computations, Texts in Applied Mathematics, vol. 59. Springer (2015)
Chung, J., Palmer, K.: A hybrid LSMR algorithm for large-scale Tikhonov regularization. SIAM J. Sci. Comput. 37, S562–S580 (2015)
Craig, E.J.: The N-step iteration procedures. J. Math. Phys. 34, 64–73 (1955)
Demmel, J.: Applied Numerical Linear Algebra. SIAM, hiladelphia (1997)
Eicke, B, Lious, A.K., Plato, R.: The instability of some gradient methods for ill-posed problems. Numer. Math. 58, 129–134 (1990)
Engl, H.W.: Regularization methods for the stable solution of inverse problems. Surveys Math. Indust. 3, 71–143 (1993)
Engl, H.W., Hanke, M., Neubauer, A.: Regularization of Inverse Problems. Kluwer Academic Publishers (2000)
Fong, D.C.L., Saunders, M.: LSMR: An iterative algorithm for sparse least-squares problems. SIAM J. Sci. Comput. 33, 2950–2971 (2011)
Gazzola, S., Hansen, P.C., Nagy, J.G.: IR tools: A MATLAB package of iterative regularization methods and large-scale test problems. Numer. Algor. 81, 773–811 (2019)
Gazzola, S., Novati, P.: Inheritance of the discrete Picard condition in Krylov subspace methods. BIT Numer. Math. 56, 893–918 (2016)
Gilyazov, S.F., Gol’dman, N.L.: Regularization of Ill-Posed Problems by Iteration Methods. Kluwer Academic Publishers, Dordrecht (2000)
Golub, G.H., O’Leary, D.P.: Some history of the conjugate gradient and Lanczos algorithms: 1948–1976. SIAM Rev 31, 50–102 (1989)
Halko, N., Martinsson, P.G., Tropp, J.A.: Finding structure with randomness: probabilistic algorithms for constructing approximate matrix decompositions. SIAM Rev. 53, 217–288 (2011)
Hanke, M.: Conjugate Gradient Type Methods for Ill-Posed Problems. Pitman Research Notes in Mathematics Series, vol. 327. Longman, Essex (1995)
Hanke, M.: On Lanczos based methods for the regularization of discrete ill-posed problems. BIT Numer. Math. 41, 1008–1018 (2001). Suppl.
Hanke, M., Hansen, P.C.: Regularization methods for large-scale problems. Surveys Math. Indust. 3, 253–315 (1993)
Hansen, P.C.: The discrete Picard condition for discrete ill-posed problems. BIT 30, 658–672 (1990)
Hansen, P.C.: Truncated singular value decomposition solutions to discrete ill-posed problems with ill-determined numerical rank. SIAM J. Sci. Statist. Comput. 11, 502–518 (1990)
Hansen, P.C.: Rank-Deficient and Discrete Ill-Posed Problems: Numerical Aspects of Linear Inversion. SIAM, Philadelphia (1998)
Hansen, P.C.: Regularization Tools version 4.0 for Matlab 7.3. Numer. Algor. 46, 189–194 (2007)
Hansen, P.C.: Discrete Inverse Problems: Insight and Algorithms. SIAM, Philadelphia (2010)
Hansen, P.C., Saxild-Hansen, M.: AIR tools–a MATLAB package of algebraic iterative reconstruction methods. J. Comput. Appl. Math. 236, 2167–2178 (2012)
Hansen, P.C., Jorgensen, J.S.: AIR Tools II: Algebraic iterative reconstruction methods, improved implementation. Numer. Algor. 79, 107–137 (2018)
Hestenes, M.R., Stiefel, E.: Methods of conjugate gradients for solving linear systems. J. Res. Nat. Bur. Stand. 49, 409–436 (1952)
Hnětynková, M.R., Kubínová, M., Plešinger, M.: Noise representation in residuals of LSQR, LSMR, and Craig regularization. Linear Algebra Appl. 533, 357–379 (2017)
Hnětynková, M.R., Plešinger, P., Strakoš, Z.: The regularizing effect of the Golub-Kahan iterative bidiagonalization and revealing the noise level in the data. BIT Numer. Math. 49, 669–696 (2009)
Hofmann, B.: Regularization for Applied Inverse and Ill-Posed Problems. Teubner, Stuttgart (1986)
Huang, Y., Jia, Z.: Some results on the regularization of LSQR for large-scale ill-posed problems. Sci. China Math. 60, 701–718 (2017)
Jia, Z.: Approximation accuracy of the Krylov subspaces for linear discrete ill-posed problems. arXiv:math.NA/1805.10132v2
Jia, Z.: The low rank approximations and Ritz values in LSQR for linear discrete ill-posed problems. arXiv:math.NA/1811.03454v1
Jia, Z.: The Krylov subspaces, low rank approximations and Ritz values in LSQR for linear discrete ill-posed problems: the multiple singular value case. In preparation (2019)
Kaipio, J., Somersalo, E.: Statistical and Computational Inverse Problems. Springer, New York (2005)
Kern, M.: Numerical Methods for Inverse Problems. Wiley (2016)
Kirsch, A.: An Introduction to the Mathematical Theory of Inverse Problems, 2nd edn. Springer, New York (2011)
Meurant, G.: The Lanczos and Conjugate Gradient Algorithms: From Theory to Finite Precision Computations. SIAM, Philadelphia (2006)
Natterer, F.: The Mathematics of Computerized Tomography. SIAM, Philadelphia (2001)
Paige, C.C., Saunders, M.A.: Solutions of sparse indefinite systems of linear equations. SIAM J. Numer. Anal. 12, 617–629 (1975)
Paige, C.C., Saunders, M.A.: LSQR: An algorithm for sparse linear equations and sparse least squares. ACM Trans. Math. Softw. 8, 43–71 (1982)
Parlett, B.N.: The Symmetric Eigenvalue Problem. SIAM, Philadelphia (1998)
Stewart, G.W., Sun, J.-G.: Matrix Perturbation Theory. Academic Press, Inc., Boston (1990)
Varah, J.M.: A practical examination of some numerical methods for linear discrete ill-posed problems. SIAM Rev. 21, 100–111 (1979)
Vogel, C.R.: Computational Methods for Inverse Problems. SIAM, Philadelphia (2002)
Wilkinson, J.H.: The Algebraic Eigenvalue Problem. Clarendon Press, Oxford (1965)
Acknowledgements
I thank two referees very much for their careful reading of the paper and for their valuable suggestions and comments, which helped me to improve the presentation.
Funding
This work was financially supported in part by the National Natural Science Foundation of China (No. 11771249).
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Jia, Z. Regularization properties of Krylov iterative solvers CGME and LSMR for linear discrete ill-posed problems with an application to truncated randomized SVDs. Numer Algor 85, 1281–1310 (2020). https://doi.org/10.1007/s11075-019-00865-w
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11075-019-00865-w
Keywords
- Discrete ill-posed
- Rank k approximations
- TSVD solution
- Semi-convergence
- Regularized solution
- Regularizing effects
- CGME
- LSMR
- LSQR