Information Technology Reference
In-Depth Information
C 4 (J
×
Theorem 2.3.8 If u
G) , we have
1
θ< 2
(i) Fo r
2
1 or for 0
and ( 2.20 ),
h 2
ε m
C(u)(h 2
su m
2
+
k)
;
1
(ii) Fo r θ
=
2 ,
h 2
ε m
C(u)(h 2
k 2 ),
su m
2
+
where the constant C(u) > 0 depends on the exact solution u and its derivatives .
We next explain the finite element method which is based on variational formu-
lations of the differential equations.
2.3.3 Finite Element Method
For the discretization with finite elements, we use the method of lines where we first
only discretize in space to obtain a system of coupled ordinary differential equations
(ODEs). In a second step, a time discretization scheme is applied to solve the ODEs.
We do not require the PDE ( 2.4 ) to hold pointwise in space but only in the varia-
tional sense. Therefore, we fix t
C 0
J and let v
(G) be a smooth test function
satisfying v(a)
0. We multiply the PDE with v , integrate with respect to
the space variable x and use integration by parts to obtain
=
v(b)
=
b
b
b
t uv d x
xx uv d x
=
fv d x,
a
a
a
x u(x,t)v(x) x = b
x = a
b
b
b
d
d t
uv d x
+
x u∂ x v d x =
fv d x.
a
a
a
=
0
Therefore, we have
b
b
b
d
d t
C 0
uv d x
+
x u∂ x v d x
=
fv d x,
v
(G).
(2.21)
a
a
a
Since C 0 (G) is not a closed subspace of L 2 (G) , we will consider test functions in
the Sobolev space H 0 (G) which is the closure of C 0
(G) in the H 1 -norm,
2
2
u
2
u
H 1 (G) =
u
L 2 (G) +
L 2 (G) .
The Sobolev space H 0 (G) consists of all continuous functions which are piecewise
differentiable and vanish at the boundary. Since ( 2.21 ) holds for all v
C 0
(G) ,
Search WWH ::




Custom Search