FAST AND FORWARD STABLE RANDOMIZED ALGORITHMS FOR LINEAR LEAST-SQUARES PROBLEMS

被引:0
作者
Epperly, Ethan N. [1 ]
机构
[1] CALTECH, Div Comp & Math Sci, Pasadena, CA 91125 USA
基金
美国国家科学基金会;
关键词
least-squares; numerical stability; randomized algorithm; sketching;
D O I
10.1137/23M1616790
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
Iterative sketching and sketch-and-precondition are randomized algorithms used for solving overdetermined linear least-squares problems. When implemented in exact arithmetic, these algorithms produce high-accuracy solutions to least-squares problems faster than standard direct methods based on QR factorization. Recently, Meier et al. demonstrated numerical instabilities in a version of sketch-and-precondition in floating point arithmetic. The work of Meier et al. raises the question, is there a randomized least-squares solver that is both fast and stable? This paper resolves this question in the affirmative by proving that iterative sketching, appropriately implemented, is forward stable. Numerical experiments confirm the theoretical findings, demonstrating that iterative sketching is stable and faster than QR-based solvers for large problem instances.
引用
收藏
页码:1782 / 1804
页数:23
相关论文
共 50 条
  • [41] Fast Measurements with MOX Sensors: A Least-Squares Approach to Blind Deconvolution
    Martinez, Dominique
    Burgues, Javier
    Marco, Santiago
    [J]. SENSORS, 2019, 19 (18)
  • [42] Stability of conjugate gradient and Lanczos methods for linear least squares problems
    Bjorck, A
    Elfving, T
    Strakos, Z
    [J]. SIAM JOURNAL ON MATRIX ANALYSIS AND APPLICATIONS, 1998, 19 (03) : 720 - 736
  • [43] Error propagation analysis of fast recursive least squares algorithms
    Arezki, M.
    Meyrueis, P.
    Benallal, A.
    Guessoum, A.
    Berkani, D.
    [J]. PROCEEDINGS OF THE NINTH IASTED INTERNATIONAL CONFERENCE ON SIGNAL AND IMAGE PROCESSING, 2007, : 101 - +
  • [44] Analysis of fast recursive least squares algorithms for adaptive filtering
    Arezki, M.
    Beneallal, A.
    Meyrueis, P.
    Guessoum, A.
    Berkani, D.
    [J]. PROCEEDINGS OF THE 11TH WSEAS INTERNATIONAL CONFERENCE ON SYSTEMS, VOL 2: SYSTEMS THEORY AND APPLICATIONS, 2007, : 473 - +
  • [45] Generalized least-squares polynomial preconditioners for symmetric indefinite linear equations
    Liang, Y
    Weston, J
    Szularz, M
    [J]. PARALLEL COMPUTING, 2002, 28 (02) : 323 - 341
  • [46] LSLQ: AN ITERATIVE METHOD FOR LINEAR LEAST-SQUARES WITH AN ERROR MINIMIZATION PROPERTY
    Estrin, Ron
    Orban, Dominique
    Saunders, Michael A.
    [J]. SIAM JOURNAL ON MATRIX ANALYSIS AND APPLICATIONS, 2019, 40 (01) : 254 - 275
  • [47] MODIFYING THE QR-DECOMPOSITION TO CONSTRAINED AND WEIGHTED LINEAR LEAST-SQUARES
    GULLIKSSON, M
    WEDIN, PA
    [J]. SIAM JOURNAL ON MATRIX ANALYSIS AND APPLICATIONS, 1992, 13 (04) : 1298 - 1313
  • [48] Pathwise least-squares estimator for linear SPDEs with additive fractional noise
    Kriz, Pavel
    Snuparkova, Jana
    [J]. ELECTRONIC JOURNAL OF STATISTICS, 2022, 16 (01): : 1561 - 1594
  • [49] Reweighted Discriminative Optimization for least-squares problems with point cloud registration
    Zhao, Yan
    Tang, Wen
    Feng, Jun
    Wan, Tao Ruan
    Xi, Long
    [J]. NEUROCOMPUTING, 2021, 464 : 48 - 71
  • [50] A least-squares approach for discretizable distance geometry problems with inexact distances
    Douglas S. Gonçalves
    [J]. Optimization Letters, 2020, 14 : 423 - 437