Abstract
The pivoted QLP decomposition, introduced by Stewart [20], represents the first two steps in an algorithm which approximates the SVD. The matrix AΠ0 is first factored as AΠ0=QR, and then the matrix R TΠ1 is factored as R TΠ1=PL T, resulting in A=QΠ1 LP TΠ0 T, with Q and P orthogonal, L lower-triangular, and Π0 and Π1 permutation matrices. Stewart noted that the diagonal elements of L approximate the singular values of A with surprising accuracy. In this paper, we provide mathematical justification for this phenomenon. If there is a gap between σ k and σ k+1, partition the matrix L into diagonal blocks L 11 and L 22 and off-diagonal block L 21, where L 11 is k-by-k. We show that the convergence of (σ j (L 11)−1−σ j −1)/σ j −1 for j=1,. . .,k, and of (σ j (L 22)−σ k+j )/σ k+j , for j=1,. . .,n−k, are all quadratic in the gap ratio σ k+1/σ k . The worst case is therefore at the gap, where the absolute errors ‖L 11 −1‖−σ k −1 and ‖L 22‖−σ k+1 are thus cubic in σ k −1 and σ k+1, respectively. One order of convergence is due to the rank-revealing pivoting in the first step; then, because of the pivoting in the first step, two more orders are achieved in the second step. Our analysis assumes that Π1=I, that is, that pivoting is done only on the first step. Although our results explain some of the properties of the pivoted QLP decomposition, they hypothesize a gap in the singular values. However, a simple example shows that the decomposition can perform well even in the absence of a gap. Thus there is more to explain, and we hope that our paper encourages others to tackle the problem. The QLP algorithm can be continued beyond the first two steps, and we make some observations concerning the asymptotic convergence. For example, we point out that repeated singular values can accelerate convergence of individual elements. This, in addition to the relative convergence to all of the singular values being quadratic in the gap ratio, further indicates that the QLP decomposition can be powerful even when the ratios between neighboring singular values are close to one.
Similar content being viewed by others
References
C.H. Bischof and C. Van Loan, The WY representation for products of householder matrices, SIAM J. Sci. Statist. Comput. 8 (1987) 2–13.
T.F. Chan, Rank revealing QR factorizations, Linear Algebra Appl. 88/89 (1987) 67–82.
S. Chandrasekaran and I.C.F. Ipsen, On rank-revealing factorisations, SIAM J. Matrix Anal. Appl. 15(2) (1994) 592–622.
S. Chandrasekaran and I.C.F. Ipsen, Analysis of a QR algorithm for computing singular values, SIAM J. Matrix Anal. Appl. 16(2) (1995) 520–535.
J.W. Demmel, Applied Numerical Linear Algebra (SIAM, Philadelphia, PA, 1997).
S. Eisenstat and I. Ipsen, Relative perturbation techniques for singular value problems, SIAM J. Numer. Anal. 32 (1995) 1972–1988.
L.V. Foster, Rank and null space calculations using matrix decomposition without column interchanges, Linear Algebra Appl. 74 (1986) 47–71.
G. Golub and C. Van Loan, Matrix Computations, 3rd ed. (Johns Hopkins Univ. Press, Baltimore, MD, 1996).
G.H. Golub, Numerical methods for solving least squares problems, Numer. Math. 7 (1965) 206–216.
S.A. Goreinov, E.E. Tyrtyshnikov and N.L. Zamarashkin, A theory of pseudoskeleton approximations, Linear Algebra Appl. 261 (1997) 1–21.
W. Kahan, Numerical linear algebra, Canadian Math. Bull. 9 (1966) 757–801.
J.C. Lagarias, Monotonicity properties of the toda flow, the QR-flow, and subspace iteration, SIAM J. Matrix Anal. Appl. 12(3) (1991) 449–462.
R. Mathias and G.W. Stewart, A block QR algorithm and the singular value decomposition, Linear Algebra Appl. 182 (1993) 91–100.
Private communication.
R.S. Schreiber and C. Van Loan, A storage-efficient WY representation for products of Householder transformations, SIAM J. Sci. Statist. Comput. 10 (1989) 52–57.
G.W. Stewart, Introduction to Matrix Computations (Academic Press, New York, 1973).
G.W. Stewart, An updating algorithm for subspace tracking, IEEE Trans. Signal Process. 40 (1992) 1535–1541.
G.W. Stewart, Updating a rank-revealing ULV decomposition, SIAM J. Matrix Anal. Appl. 14 (1993) 494–499.
G.W. Stewart, Matrix Algorithms, Vol. 1: Basic Decompositions (SIAM, Philadelphia, PA, 1998).
G.W. Stewart, The QLP approximation to the singular value decomposition, SIAM J. Sci. Comput. 20(4) (1999) 1336–1348.
L.N. Trefethen and D. Bau, III, Numerical Linear Algebra (SIAM, Philadelphia, PA, 1997).
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Huckaby, D.A., Chan, T.F. On the Convergence of Stewart's QLP Algorithm for Approximating the SVD. Numerical Algorithms 32, 287–316 (2003). https://doi.org/10.1023/A:1024082314087
Issue Date:
DOI: https://doi.org/10.1023/A:1024082314087