Sie sind auf Seite 1von 2

Algorithmen zu den Abstiegsverfahren (Kap.

6)
a) Verfahren des steilsten Abstiegs
x
(0)
gegeben
r
(k)
= b Ax
(k)

k
=
(r
(k)
)
T
r
(k)
(r
(k)
)
T
Ar
(k)
x
(k+1)
= x
(k)
+
k
r
(k)

k = 0, 1, . . .
b) cgVerfahren (teiloptimiert; nur 1 MatrixVektorMultiplikation!)
x
(0)
gegeben, r
(0)
= b Ax
(0)
= s
(0)

k
=
(r
(k)
)
T
r
(k)
(s
(k)
)
T
(As
(k)
)
x
(k+1)
= x
(k)
+
k
s
(k)
r
(k+1)
= r
(k)

k
(As
(k)
)

k
=
(r
(k+1)
)
T
r
(k+1)
(r
(k)
)
T
r
(k)
s
(k+1)
= r
(k+1)
+
k
s
(k)

k = 0, 1, . . .
Abbruch mit Losung x

= x
(k)
, falls r
(k)
= 0.
Beachte:
(i) r
(k+1)
= b Ax
(k+1)
= b A(x
(k)
+
k
s
(k)
) = r
(k)
As
(k)
, wie oben
(ii) (r
(k)
)
T
s
(k)
= (r
(k)
)
T
(r
(k)
+
k1
s
(k1)
) = (r
(k)
)
T
r
(k)
(iii) As
(k)
und (r
(k)
)
T
r
(k)
sind zwischenzuspeichern.
1
c) cgVerfahren mit Vorkonditionierung
x
(0)
gegeben, r
(0)
= b Ax
(0)
solve Mr
(k)
v
= r
(k)
if k = 0
s
(0)
v
= r
(0)
v
else

k1
=
(r
(k)
)
T
r
(k)
v
(r
(k1)
)
T
r
(k1)
v
s
(k)
v
= r
(k)
v
+

k1
s
(k1)
v
end

k
=
(r
(k)
)
T
r
(k)
v
(s
(k)
v
)
T
(As
(k)
v
)
x
(k+1)
= x
(k)
+
k
s
(k)
v
r
(k+1)
= r
(k)

k
(As
(k)
v
)

k = 0, 1, . . .
Abbruch mit Losung x

= x
(k)
, falls r
(k)
= 0.
Beachte:
As
(k)
v
und (r
(k)
)
T
r
(k)
v
sind zwischenzuspeichern.
2