3.2 Convergence Analysis for__ Iterative Methods 数值分析课件.pdf

3.2 Convergence Analysis for Iterative Method Theorem 3.2.1 (0) n (k) ∞ For any x ∈ R , the sequence {x }k=0 defined by x(k) = Bx(k−1) + d, for k = 1, 2, 3, · · · , (3.2.1) converges to the unique solution x∗ of x = Bx + d if and only if ρ(B) < 1. Theorem 3.2.2 If B = q < 1 for any natural matrix norm, then for any Home Page (0) n (k) ∞ (k) (k−1) x ∈ R , the sequence {x }k=0 defined by x = Bx + d converges Title Page to the unique solution x∗ of x = Bx + d. Moreover the following error bounds hold for k = 1, 2, 3, · · · : ∗ (k) q (k) (k−1) (a) x − x ≤ x − x ; (3.2.2) Page 1 of 12 1 − q Go Back k (b) x∗ − x(k) ≤ q x(1) − x(0) . (3.2.3) Full Screen 1 − q Close Proof See the textbook pp. 051-052. Quit 3.2 SSS“““{{{ÂÂÂñññ555©©©ÛÛÛ ½½½nnn3.2.1 é?¿ÐŠx(0) ∈ Rn §dS“‚ª x(k) = Bx(k−1) + d, k = 1, 2, 3, · · · , (3.2.1) )S{x(k) }∞ ∗ k=0Âñ§|x = Bx + d ˜)x ¿