一种有效的新预条件方法.pdfVIP

2008,28A(1):024–034  ∗ ( 710049) Z - Jacobi Gauss-Seidel Æ  Z - MR(2000) 65F10 O241.6 A 1003-3998(2008)01-024-11 1 Ax = b, A ∈ Rn ×n (1) A Æ P , (1) P A Z - Æ [2−4] . [4] Æ Jacobi Gauss-Seidel [4], A Z - (Z - ℄), A A = I − L − U ∈ Rn ×n , I A −L −U A Ω = diag(ω , ω , , ω ), 0 ≤ ω ≤ 1, i = 1, 2, , n, 1 2 n i Γ = diag(γ , γ , , γ ), 0 ≤ γ ≤ 1, i = 1, 2, , n, 1 2 n i [3] (I + U ) (1) A [4] 0 ≤ β ≤ 1, (I + βU ) (1) A β A Z - 0 ≤ β ≤ 1,(I + βU )A Z - ℄ I + ΩU + ΓL, 2005-11-05; 2007-10-08 E-mail: jcli@mail.xjtu.edu.cn ∗ (2007A16) No.1 25 A AΓ = (I + ΩU + ΓL)A Ω Z - Jacobi Gauss-Seidel ℄ ℄ 1.1 A Z -  1 −0.8 −0.1 −0.01 −0.001    −0.1 −1 −0.6 −0.2 −0.1        A = −0.4 −0.1 1 −0.51 −0.1 ,       −0.1 −0.3 −0.2 1 −0.1 −0.2 −0.2 −0.2 −0.4 1  L U  0.87 −0.0132 −0.4822 −0.2114 −0.091 