Chapter 3 Introduction To Sobolev Spaces

Download as pdf or txt
Download as pdf or txt
You are on page 1of 52
 
Chapter 3
Introduction to SobolevSpaces
Remark 3.1
 Contents.
 Sobolev spaces are the basis of the theory of weak orvariational forms of partial differential equations. A very popular approach fordiscretizing partial differential equations, the finite element method, is based onvariational forms. In this chapter, a short introduction into Sobolev spaces will begiven. Recommended literature are the books Adams (1975); Adams and Fournier(2003) and Evans (2010).
 
3.1 Elementary Inequalities
Lemma 3.2 Inequality for strictly monotonically increasing function.
 Let 
 :
 R
+
 ∪{
0
}
 R
 be a continuous and strictly monotonically increasing function with 
 
(0) = 0
 and 
 
(
x
)
 →
 for 
 x
 →
. Then, for all 
 a,b
 ∈
R
+
 ∪{
0
}
 it is 
ab
 ≤
 
 a
0
(
x
)
 dx
+
 
 b
0
1
(
y
)
 dy,
where 
 
1
(
y
)
 is the inverse of 
 
(
x
)
.
Proof:
 Since
 f 
(
x
) is strictly monotonically increasing, the inverse function exists.The proof is based on a geometric argument, see Figure 3.1.
Figure 3.1: Sketch to the proof of Lemma 3.2.
Consider the interval (0
,a
) on the
 x
-axis and the interval (0
,b
) on the
 y
-axis. Then, thearea of the corresponding rectangle is given by
 ab
,
 
a
0
 
(
x
)
 dx
 is the area below the curve,and
 
b
0
 f 
1
(
y
)
 dy
 is the area between the positive
 y
-axis and the curve. From Figure 3.1,the inequality follows immediately. The equal sign holds only iff 
 f 
(
a
) =
 b
.
36
 
Remark 3.3
 Young’s inequality.
 Young’s inequality
ab
 ≤
 ε
2
a
2
+ 12
εb
2
 a,b,ε
 ∈
R
+
 (3.1)follows from Lemma 3.2 with
 
(
x
) =
 εx
,
 
1
(
y
) =
 ε
1
y
. It is also possible toderive this inequality from the binomial theorem. For proving the generalized Younginequality
ab
 ≤
 ε
 p
 p a
 p
+ 1
q
b
q
,
 
 a,b,ε
 ∈
R
+
 (3.2)with
 p
1
+
 q 
1
= 1
,p,q 
 ∈
 (1
,
), one chooses
 f 
(
x
) =
 x
 p
1
,
 f 
1
(
y
) =
 y
1
/
(
 p
1)
andapplies Lemma 3.2 with intervals where the upper bounds are given by
 εa
 and
 ε
1
b
.
Remark 3.4
 Cauchy–Schwarz inequality.
 The Cauchy
1
–Schwarz
2
inequality (forvectors, for sums)
|
(
x
,
y
)
| ≤
x
2
y
2
 
 x
,
y
 ∈
R
n
,
 (3.3)where (
·
,
·
) is the Euclidean product and
 ·
2
 the Euclidean norm, is well known.One can prove this inequality with the help of Young’s inequality.First, it is clear that the Cauchy–Schwarz inequality is correct if one of thevectors is the zero vector. Now, let
 x
,
y
 with
 
x
2
 =
 
y
2
 = 1. One obtains withthe triangle inequality and Young’s inequality (3.1)
|
(
x
,
y
)
|
 =
n
i
=1
x
i
y
i
n
i
=1
|
x
i
||
y
i
| ≤
 12
n
i
=1
|
x
i
|
2
+ 12
n
i
=1
|
y
i
|
2
= 1
.
Hence, the Cauchy–Schwarz inequality is correct for
 x
,
y
. Last, one considers arbi-trary vectors ˜
x
 
=
 0
,
 ˜
y
 
=
 0
. Now, one can utilize the homogeneity of the Cauchy–Schwarz inequality. From the validity of the Cauchy–Schwarz inequality for
 x
 and
y
, one obtains by a scaling argument
(
˜
x
12
 ˜
x
    
x
,
˜
y
12
 ˜
y
    
y
)
 1Both vectors
 x
,
y
 have the Euclidean norm 1, hence1
˜
x
2
˜
y
2
|
(˜
x
,
 ˜
y
)
| ≤
 1
 |
(˜
x
,
 ˜
y
)
| ≤
˜
x
2
˜
y
2
.
The generalized Cauchy–Schwarz inequality or H¨older inequality
|
(
x
,
y
)
| ≤
 n
i
=1
|
x
i
|
 p
1
/p
 n
i
=1
|
y
i
|
q
1
/q
with
 p
1
+
 q 
1
= 1
,p,q 
 ∈
 (1
,
), can be proved in the same way with the help of the generalized Young inequality.
 
Definition 3.5
 Lebesgue spaces.
 The space of functions which are Lebesgueintegrable on Ω to the power of 
 p
 ∈
 [1
,
) is denoted by
L
 p
(Ω) =
 :
 
|
|
 p
(
x
)
 d
x
 <
 ∞
,
1
Augustin Louis Cauchy (1789 – 1857)
2
Hermann Amandus Schwarz (1843 – 1921)
37
5
 
which is equipped with the norm
L
p
(
x
)
 =
 
|
|
 p
(
x
)
 d
x
1
/p
.
For
 p
 =
 ∞
, this space is
L
(Ω) =
 {
 :
 |
(
x
)
|
 <
 ∞
 almost everywhere in Ω
}
with the norm
L
(Ω)
 = ess sup
x
|
(
x
)
|
.
Lemma 3.6 older’s inequality.
 Let 
 p
1
+
1
= 1
,p,q 
 ∈
 [1
,
]
. If 
 u
 ∈
 L
 p
(Ω)
and 
 v
 ∈
 L
q
(Ω)
, then it is 
 uv
 ∈
 L
1
(Ω)
 and it holds that 
uv
L
1
(Ω)
 ≤
u
L
p
(Ω)
v
L
q
(Ω)
.
 (3.4)
If 
 p
 =
 q 
 = 2
, then this inequality is also known as Cauchy–Schwarz inequality 
uv
L
1
(Ω)
 ≤
u
L
2
(Ω)
v
L
2
(Ω)
.
 (3.5)
Proof:
 p,
 
 (1
,
). First, one has to show that
 |
uv
(
x
)
|
 can be estimated fromabove by an integrable function. Setting in the generalized Young inequality (3.2)
 ε
 = 1,
a
 =
 |
u
(
x
)
|
, and
 b
 =
 |
v
(
x
)
|
 gives
|
u
(
x
)
v
(
x
)
|
 1
 p
 |
u
(
x
)
|
p
+ 1
 |
v
(
x
)
|
q
.
Since the right hand side of this inequality is integrable, by assumption, it follows that
uv
 ∈
 L
1
(Ω). In addition, H¨older’s inequality is proved for the case
u
L
p
(Ω)
 =
 
v
L
q
(Ω)
 = 1using this inequality
 
|
u
(
x
)
v
(
x
)
|
 d
x
 ≤
 1
 p
 
|
u
(
x
)
|
p
d
x
+ 1
 
|
v
(
x
)
|
q
d
x
 = 1
.
The general inequality follows, for the case that both functions do not vanish almosteverywhere, with the same homogeneity argument as used for proving the Cauchy–Schwarzinequality of sums. In the case that one of the functions vanishes almost everywhere, (3.4)is trivially satisfied.
 p
 = 1
,
 =
 ∞
. It is
 
|
u
(
x
)
v
(
x
)
|
 d
x
 ≤
 
|
u
(
x
)
|
ess sup
x
|
v
(
x
)
|
 d
x
 =
 
u
L
1
(Ω)
v
L
(Ω)
 .
3.2 Weak Derivative and Distributions
Remark 3.7
 Contents.
 This section introduces a generalization of the derivativewhich is needed for the definition of weak or variational problems. For an introduc-tion to the topic of this section, see, e.g., Haroske and Triebel (2008)Let Ω
 ⊂
 R
d
be a domain with boundary Γ =
 ∂ 
Ω,
 d
 ∈
 N
,
 
=
 ∅
. A domain isalways an open set.
 
Definition 3.8 The space
 C 
0
 (Ω)
.
 The space of infinitely often differentiable realfunctions with compact (closed and bounded) support in Ω is denoted by
 C 
0
 (Ω)
0
 (Ω) =
 {
v
 :
 v
 ∈
 C 
(Ω)
,
 supp(
v
)
 ⊂
 Ω
}
,
wheresupp(
v
) =
 {
x
 ∈
 Ω :
 v
(
x
)
 
= 0
}
.
38
5
 
Definition 3.9 Convergence in
 
0
 (Ω)
.
 The sequence of functions
 {
φ
n
(
x
)
}
n
=1
,
φ
n
 ∈
 C 
0
 (Ω) for all
 n
, is said to convergence to the zero functions if and only if a)
 
 ⊂
 Ω
,
 compact (closed and bounded) with supp(
φ
n
)
 ⊂
 K 
 for all
 n
,b)
 D
α
φ
n
(
x
)
 
 0 for
 n
 
 on
 
 for all multi-indices
 α
 = (
α
1
,...,α
d
),
 |
α
|
 =
α
1
 +
...
+
α
d
.It islim
n
→∞
φ
n
(
x
) =
 φ
(
x
)
 
 lim
n
→∞
(
φ
n
(
x
)
φ
(
x
)) = 0
.
Definition 3.10 Weak derivative.
 Let
 f,
 
 L
1loc
(Ω). (
L
1loc
(Ω): for each com-pact subset Ω
 Ω it holds
 
|
u
(
x
)
|
 d
x
 <
 ∞
 u
 ∈
 L
1loc
(Ω)
.
)If for all functions
 g
 ∈
 C 
0
 (Ω) it holds that
 
(
x
)
g
(
x
)
 d
x
 = (
1)
|
α
|
 
(
x
)
D
α
g
(
x
)
 d
x
,
then
 F 
(
x
) is called weak derivative of 
 f 
(
x
) with respect to the multi-index
 α
.
 
Remark 3.11
 On the weak derivative.
 One uses the same notations for the derivative as in the classical case :
 
(
x
) =
D
α
(
x
).
 If 
 f 
(
x
) is classically differentiable on Ω, then the classical derivative is also theweak derivative.
 The assumptions on
 f 
(
x
) and
 F 
(
x
) are such that the integrals in the definitionof the weak derivative are well defined. In particular, since the test functionsvanish in a neighborhood of the boundary, the behavior of 
 
(
x
) and
 
(
x
) if 
 x
approaches the boundary is not of importance.
 The main aspect of the weak derivative is due to the fact that the (Lebesgue)integral is not influenced from the values of the functions on a set of (Lebesgue)measure zero. Hence, the weak derivative is uniquely defined only up to a setof measure zero. It follows that
 f 
(
x
) might be not classically differentiable on aset of measure zero, e.g., in a point, but it can still be weakly differentiable.
 The weak derivative is uniquely determined, in the sense described above.
Example 3.12
 Weak derivative.
 The weak derivative of the function
 f 
(
x
) =
 |
x
|
 is
(
x
) =
1
 x <
 00
 x
 = 01
 x >
 0In
 x
 = 0, one can use also any other real number. The proof of this statementfollows directly from the definition and it is left as an exercise.
 
Definition 3.13 Distribution.
 A continuous linear functional defined on
 C 
0
 (Ω)is called distribution. The set of all distributions is denoted by (
0
 (Ω))
.Let
 u
 ∈
 C 
0
 (Ω) and
 ψ
 
 (
0
 (Ω))
, then the following notation is used for theapplication of the distribution to the function
ψ
(
u
(
x
)) =
 
ψ,u
 ∈
R
.
39
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5

Reward Your Curiosity

Everything you want to read.
Anytime. Anywhere. Any device.
No Commitment. Cancel anytime.
576648e32a3d8b82ca71961b7a986505