The randomized Kaczmarz algorithm with the probability distribution depending on the angle

被引:4
作者
He, Songnian [1 ,2 ]
Dong, Qiao-Li [1 ,2 ]
Li, Xiaoxiao [2 ]
机构
[1] Civil Aviat Univ China, Tianjin Key Lab Adv Signal Proc, Tianjin 300300, Peoples R China
[2] Civil Aviat Univ China, Coll Sci, Tianjin 300300, Peoples R China
关键词
Linear system; Projection; Randomized Kaczmarz algorithm; Probability distribution; Accelerated scheme;
D O I
10.1007/s11075-022-01422-8
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
In this paper we propose a new probability distribution for the randomized Kaczmarz (RK) algorithm where each row of the coefficient matrix is selected in the current iteration with the probability proportional to the square of the sine of the angle between it and the chosen row in the previous iteration. This probability distribution is helpful to accelerate the convergence of the RK algorithm. We obtain the linear convergence rate estimation in expectation for the RK algorithm with the new probability distribution (RKn), which is different from the existing results. In order to avoid the calculation and storage of probability matrix when solving the large-scale linear systems, one practical probability distribution is also proposed. Furthermore, an acceleration for the RKn algorithm and its simplified version are introduced respectively, whose core is to replace the projection onto one hyperplane with that onto the intersection of two hyperplanes. The accelerated schemes are proved to have faster convergence rates. Finally, numerical experiments are provided to illustrate the performance of the proposed algorithms by comparing with the RK algorithm and Motzkin's algorithm.
引用
收藏
页码:415 / 440
页数:26
相关论文
共 14 条
  • [1] ON GREEDY RANDOMIZED KACZMARZ METHOD FOR SOLVING LARGE SPARSE LINEAR SYSTEMS
    Bai, Zhong-Zhi
    Wu, Wen-Ting
    [J]. SIAM JOURNAL ON SCIENTIFIC COMPUTING, 2018, 40 (01) : A592 - A606
  • [2] Bauschke HH, 2011, CMS BOOKS MATH, P1, DOI 10.1007/978-1-4419-9467-7
  • [3] Byrne CL, 2014, MONOGR RES NOTES MAT, P1
  • [4] On the Randomized Kaczmarz Algorithm
    Dai, Liang
    Soltanalian, Mojtaba
    Pelckmans, Kristiaan
    [J]. IEEE SIGNAL PROCESSING LETTERS, 2014, 21 (03) : 330 - 333
  • [5] Algorithm 915, SuiteSparseQR: Multifrontal Multithreaded Rank-Revealing Sparse QR Factorization
    Davis, Timothy A.
    [J]. ACM TRANSACTIONS ON MATHEMATICAL SOFTWARE, 2011, 38 (01):
  • [6] Goebel K., 1984, UNIFORM CONVEXITY HY
  • [7] Gower R.M., 2015, ARXIV
  • [8] On Motzkin's method for inconsistent linear systems
    Haddock, Jamie
    Needell, Deanna
    [J]. BIT NUMERICAL MATHEMATICS, 2019, 59 (02) : 387 - 401
  • [9] Realization of the hybrid method for Mann iterations
    He, Songnian
    Yang, Caiping
    Duan, Peichao
    [J]. APPLIED MATHEMATICS AND COMPUTATION, 2010, 217 (08) : 4239 - 4247
  • [10] Kaczmarz S., 1937, B ACAD POLON SCI L A, V35, P355