248 CHAP. 20 Numeric Linear Algebra
Table 20.2 Gauss-Seidel Iteration
ALGORITHM GAUSS-SEIDEL (A, b, x, ¢ N) |
‘This algorithm computes solution x ofthe system Ax = b given an inital approximation
2 where A= [ay] is ann Xn mati with ay # Oj = Io" sm
INPUT: A, b, intial approximation x, tolerance € > 0, maximum number
of iterations N |
OUTPUT: Approximate solution x™” = [4] or failure message that" does
not satis the tolerance eonltion
Form = 0,-++,.N — L,do:
Forj = 1-1 do:
1, oe °
1 gen t (Sanit? 3 ager) |
End |
2 Imax aj"? = 2$] < ethon OUTPUT X*?, Stop |
End
OUTPUT: “No solution satisfying the tolerance condition obtained after N
iteration steps.” Stop
[Procedure completed unsuccessfully)
End GAUISS-SEIDEL
[Procedure completed successful] |
Convergence and Matrix Norms
‘An iteration method for solving Ax = b is said to converge for an initial x if the
corresponding iterative sequence x®, x", x, -«- converges to a solution of the given
system, Convergence depends on the relation between x” and x". To get this relation
for the Gauss-Seidel method, we use (6). We first have
d+ Ex
b- ux"
and by multiplying by (I + 1)? from the left,
O mse 40+ where
The Gauss-Seidel iteration converges for every x®? if and only if all the eigenvalues
(Gee. 8.1) of the “iteration matrix” C = ey] have absolute value less than 1. (Proof in
Ref. {ES}, p. 191, listed in App. 1.)
CAUTION! If you want to get C, first divide the rows of A by a, to have main diagonal
1, ++, 1. Ifthe spectral radius of C (= maximum of those absolute values) is small,
then the convergence is rapid
‘Sufficient Convergence Condition, A sufficient condition for convergence is,
6 lley <1