Smoothed analysis of the condition numbers and growth factors of matrices

被引:102
作者
Sankar, Arvind [1 ]
Spielman, Daniel A.
Teng, Shang-Hua
机构
[1] MIT, Dept Math, Cambridge, MA 02139 USA
[2] Yale Univ, Dept Comp Sci, New Haven, CT 06250 USA
[3] Boston Univ, Dept Comp Sci, Boston, MA 02215 USA
[4] Akamai Technol Inc, Cambridge, MA 02142 USA
关键词
smoothed analysis; condition number; Gaussian elimination; growth factor;
D O I
10.1137/S0895479803436202
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
Let (A) over bar be an arbitrary matrix and let A be a slight random perturbation of (A) over bar. We prove that it is unlikely that A has a large condition number. Using this result, we prove that it is unlikely that A has large growth factor under Gaussian elimination without pivoting. By combining these results, we show that the smoothed precision necessary to solve Ax = b, for any b, using Gaussian elimination without pivoting is logarithmic. Moreover, when (A) over bar is an all-zero square matrix, our results significantly improve the average-case analysis of Gaussian elimination without pivoting performed by Yeung and Chan (SIAM J. Matrix Anal. Appl., 18 (1997), pp. 499-517).
引用
收藏
页码:446 / 476
页数:31
相关论文
共 25 条
  • [1] Anderson E, 1999, LAPACK USERS GUIDE
  • [2] Blum A, 2002, SIAM PROC S, P905
  • [3] BLUM L, 1989, P COMPL SYST SUMM SC, P1
  • [4] Davidson KR, 2001, HANDBOOK OF THE GEOMETRY OF BANACH SPACES, VOL 1, P317, DOI 10.1016/S1874-5849(01)80010-3
  • [5] DEMMEL JW, 1988, MATH COMPUT, P449
  • [6] Demmel JW., 1997, APPL NUMERICAL LINEA
  • [7] DUNAGAN J, 2003, SMOOTHED ANAL RENEGA
  • [8] EIGENVALUES AND CONDITION NUMBERS OF RANDOM MATRICES
    EDELMAN, A
    [J]. SIAM JOURNAL ON MATRIX ANALYSIS AND APPLICATIONS, 1988, 9 (04) : 543 - 560
  • [9] EDELMAN A, 1992, NATO ASI SER, P365
  • [10] Golub G.H., 1983, Johns Hopkins series in the mathematical sciences