Beruflich Dokumente
Kultur Dokumente
Matrix Equations
Convergence Properties
Summary
Dr. Johnson
School of Mathematics
Semester 1 2008
O UTLINE
1 R EVIEW
2 M ATRIX E QUATIONS
Rearranging the Matrix
Iteration Matrix Equations
3 C ONVERGENCE P ROPERTIES
The Iteration Matrix
Properties of Convergence
Further Convergence Properties
4 S UMMARY
O UTLINE
1 R EVIEW
2 M ATRIX E QUATIONS
Rearranging the Matrix
Iteration Matrix Equations
3 C ONVERGENCE P ROPERTIES
The Iteration Matrix
Properties of Convergence
Further Convergence Properties
4 S UMMARY
O UTLINE
1 R EVIEW
2 M ATRIX E QUATIONS
Rearranging the Matrix
Iteration Matrix Equations
3 C ONVERGENCE P ROPERTIES
The Iteration Matrix
Properties of Convergence
Further Convergence Properties
4 S UMMARY
O UTLINE
1 R EVIEW
2 M ATRIX E QUATIONS
Rearranging the Matrix
Iteration Matrix Equations
3 C ONVERGENCE P ROPERTIES
The Iteration Matrix
Properties of Convergence
Further Convergence Properties
4 S UMMARY
R ELAXATION M ETHODS
q+1 q ∗
wi,j = (1 − ω )wi,j + ωwi,j
R ELAXATION M ETHODS
q+1 q ∗
wi,j = (1 − ω )wi,j + ωwi,j
O UTLINE
1 R EVIEW
2 M ATRIX E QUATIONS
Rearranging the Matrix
Iteration Matrix Equations
3 C ONVERGENCE P ROPERTIES
The Iteration Matrix
Properties of Convergence
Further Convergence Properties
4 S UMMARY
M ATRIX E QUATION
Ax = b
where A = (ai,j ) is an (n × n) matrix, and x, b are (n × 1)
column vectors.
M ATRIX E QUATION
M ATRIX E QUATION
0 a1,2 . . .
0 0 a2,3
0 0 a3,4
..
U= 0 .
..
..
0 . .
0 0 an−1,n
0 0
M ATRIX E QUATION
0 0
a2,1 0 0
..
. a3,2 0 0
L=
..
.
..
.
an−1,n−2 0 0
... an,n−1 0
M ATRIX E QUATION
Ax = b
as
Lx + Dx + Ux = b.
O UTLINE
1 R EVIEW
2 M ATRIX E QUATIONS
Rearranging the Matrix
Iteration Matrix Equations
3 C ONVERGENCE P ROPERTIES
The Iteration Matrix
Properties of Convergence
Further Convergence Properties
4 S UMMARY
I TERATIVE METHODS
I TERATIVE METHODS
O UTLINE
1 R EVIEW
2 M ATRIX E QUATIONS
Rearranging the Matrix
Iteration Matrix Equations
3 C ONVERGENCE P ROPERTIES
The Iteration Matrix
Properties of Convergence
Further Convergence Properties
4 S UMMARY
x(k+1) = Px(k) + Q,
P = PJ = D−1 (−L − U )
P = PG = (D + L)−1 (−U ).
x(k+1) = Px(k) + Q,
P = PJ = D−1 (−L − U )
P = PG = (D + L)−1 (−U ).
O UTLINE
1 R EVIEW
2 M ATRIX E QUATIONS
Rearranging the Matrix
Iteration Matrix Equations
3 C ONVERGENCE P ROPERTIES
The Iteration Matrix
Properties of Convergence
Further Convergence Properties
4 S UMMARY
E RROR V ECTORS
x = Px + Q.
x(k ) = x + e(k )
D IMINISHING ERRORS
||Pk || → 0 as k → ∞,
Since
||Pk || = ||P||k
||P|| < 1.
O UTLINE
1 R EVIEW
2 M ATRIX E QUATIONS
Rearranging the Matrix
Iteration Matrix Equations
3 C ONVERGENCE P ROPERTIES
The Iteration Matrix
Properties of Convergence
Further Convergence Properties
4 S UMMARY
S PECTRAL R ADIUS
||P|| < 1
C ONVERGENCE R ATE
Note also that for large k
||e(k+1) || = ρ||e(k) ||
ρq < ǫ
giving
log ǫ
q ≤ qd = .
log ρ
C ONVERGENCE R ATE
Note also that for large k
||e(k+1) || = ρ||e(k) ||
ρq < ǫ
giving
log ǫ
q ≤ qd = .
log ρ
C ONVERGENCE R ATE
Note also that for large k
||e(k+1) || = ρ||e(k) ||
ρq < ǫ
giving
log ǫ
q ≤ qd = .
log ρ
cos πn + β2 cos m
¡ ¢ ¡π¢
ρ = ρ ( PJ ) = ,
1 + β2
C ONVERGENCE P ROPERTIES
log ǫ 2n2
qd = ³ ´ = − 2 log ǫ
log 1 − 2nπ2 π
2
n2
qd = − log ǫ
π2
Gauss-Seidel converges twice as fast as Jacobi.
C ONVERGENCE P ROPERTIES
log ǫ 2n2
qd = ³ ´ = − 2 log ǫ
log 1 − 2nπ2 π
2
n2
qd = − log ǫ
π2
Gauss-Seidel converges twice as fast as Jacobi.
O PTIMAL SOR
(λ + ω − 1)2 = λω 2 µ2
2
ωb = p
1+ 1 − [ρ(PJ )]2
O PTIMAL SOR
(λ + ω − 1)2 = λω 2 µ2
2
ωb = p
1+ 1 − [ρ(PJ )]2
1 − sin πn
ρ=
1 + sin πn
giving
n
qd = − log ǫ.
2π
giving
n
qd = − √ log ǫ.
2 2π
√
Line SOR converges 2 times faster than point SOR for the
model problem.
When choosing a method we must decide whether faster
convergence compensates for extra computations required.
Dr. Johnson MATH65241
Review
The Iteration Matrix
Matrix Equations
Properties of Convergence
Convergence Properties
Further Convergence Properties
Summary
giving
n
qd = − √ log ǫ.
2 2π
√
Line SOR converges 2 times faster than point SOR for the
model problem.
When choosing a method we must decide whether faster
convergence compensates for extra computations required.
Dr. Johnson MATH65241
Review
Matrix Equations
Convergence Properties
Summary
Lx + Dx + Ux = b.
x(k+1) = Px(k) + Q,
||P|| < 1.