PDE III (M597J Fall 2004) Lecture Notes

Yuxi Zheng
Department of Mathematics
The Pennsylvania State University
University Park, PA 16802

August 31, 2004

Abstract

Chapter 1. Spaces of Functions: Hölder continuous functions; Locally Hölder continu-

ous functions; Nondimensionalized norms and seminorms; Interior seminorms and norms;
Partially interior norms; Intermediate norms (Gilbarg-Hömander’s). References [2][1].

1 Chapter 1: Spaces of functions

We find in many key steps in the theory of partial differential equations that we need fine
properties of functions. One example is the smoothness of the Newtonian potential w for a
continuous function f . Hölder continuity of f transmits to Hölder continuity of the second-
order derivatives of w. Simply being continuous does not imply continuity of D2 w, however.
Another example is the numerical approximation of smooth domains with rectangular grids,
thus corners are produced, and harmonic functions at a corner x0 typically exihibit behavior
of the type w(x) = |x − x0 |α for some real non-integral number α. The Hölder spaces and their
subspaces are handy in catching these properties.

Notations: Rn : Euclidean n-space. A point x = (x1 , x2 , · · · , xn ). Norm |x| = ( ni=1 x2i )1/2 .
P

Rn+ := {x ∈ Rn |xn > 0}: open upper half-space in Rn .

Ω: A proper open subset of Rn , not necessarily bounded.

B(y): An open ball in Rn with center y;

Br (y): An open ball in Rn with radius r centered at y;

C 0 (Ω): The set of continuous functions on Ω;

C 0 (Ω): The set of continuous functions on Ω;

C k (Ω): The set of all functions having all derivatives of order ≤ k continuous on Ω (k is a
nonnegative integer or infinity);

1
 C k (Ω): The set of all functions in C k (Ω) all of whose deriatives of order ≤ k have continuous
extensions to Ω.

supp u is the support of u, the closure of the set on which u 6= 0.

Cck (Ω): The set of functions in C k (Ω) with compact support in Ω.

See [2], pp. 9–10, for more notations.

1.1 Hölder spaces

Pointwise Hölder continuity. Let f be a function defined on a bounded set D ⊂ Rn . Let

x0 ∈ D, and 0 < α ≤ 1. We say that f is Hölder continuous with exponent α at x0 if the
quantity
|f (x) − f (x0 )|
[f ]α,x0 := sup (1)
D |x − x0 |α
is finite. We call [f ]α,x0 the α-Hölder coefficient of f at x0 with respect to D.

Example 1: The function f (x) = |x|2/3 on B1 (0) is Hölder continuous with exponent 2/3
at x = 0.

Uniform Hölder continuity. Let f be a function defined on any set D ⊂ Rn . Let

0 < α ≤ 1. We call f uniformly Hölder continuous with exponent α in D if the quantity

|f (x) − f (y)|
[f ]α,D := sup (2)
x6=y∈D |x − y|α

is finite.

Local Hölder continuity. Let f be a function defined on any set D ⊂ Rn . Let 0 < α ≤ 1.
We call f locally Hölder continuous with exponent α in D if f is uniformly Hölder continuous
with exponent α on any compact subsets of D. Local and uniform Hölder are equivalent on
compact sets D. A bounded, locally Hölder continuous function is pointwise Hölder continuous
in D.

Let Ω be an open set in Rn , 0 < α ≤ 1, and k a nonnegative integer. The (uniform)

Hölder spaces C k,α (Ω) consist of functions whose k−th order derivatives are uniformly Hölder
continuous with exponent α in Ω. The (local) Hölder spaces C k,α (Ω) consist of functions whose
k−th order derivatives are locally Hölder continuous with exponent α in Ω.

Hölder continuity may be rgarded as fractional differentiability. The two exponents (k, α)
can be combined to form one k + α. In the reverse splitting, we always choose 0 < α ≤ 1. This
practice is slightly different from Gilbarg [2].

We can provide norms for (uniform) Hölder spaces C k,α (Ω).

2
 We define the seminorms
[u]k,0;Ω = |Dk u|0;Ω = sup|β|=k supΩ |Dβ u|, k = 0, 1, 2, · · ·
(3)
[u]k,α;Ω = [Dk u]α;Ω = sup|β|=k [Dβ u]α;Ω ,

and the norms

Pk Pk j u|
kukC k (Ω) = |u|k;Ω = |u|k,0;Ω = j=0 [u]j,0;Ω = j=0 |D 0;Ω ,
(4)
kukC k,α (Ω) = |u|k,α;Ω = |u|k;Ω + [u]k,α;Ω = |u|k;Ω + [Dk u]α;Ω

on the spaces C k (Ω) and C k,α (Ω) respectively. They are Banach spaces. We note that it does
not make any difference whether we use Ω or its closure: The supremum taken in the definition
blurs distinction. One needs to use the concept of local spaces (i.e., mention the word “local”)
to make a distinction.

1.2 Non-dimensionalized norms

It is sometimes illuminating to use the equivalent, but non-dimensionalized norms of C k (Ω)

and C k,α (Ω). If Ω is bounded with d = diam Ω, we let
Pk Pk
kuk0C k (Ω) = |u|0k;Ω = j=0 d
j [u]
j,0;Ω = j=0 d
j |D j u|
0;Ω ,
(5)
kuk0C k,α (Ω) = |u|0k,α;Ω = |u|0k;Ω + dk+α [u]k,α;Ω = |u|0k;Ω + dk+α [Dk u]α;Ω .

Product rule. Non-dimensional norms are neat to present theorems. For example, the
product rule for Hölder continuous functions is

kuvk0C γ (Ω) ≤ kuk0C α (Ω) kvk0C β (Ω) ;

(6)
kuvkC γ (Ω) ≤ max(1, dα+β−2γ )kukC α (Ω) kvkC β (Ω) ,

where γ = min(α, β).

0 0
Inclusion rule For domains of interest of our class, the inclusion C k ,α (Ω) ⊂ C k,α (Ω) will
hold whenever k + α < k 0 + α0 .

See [2], Sect 4.1, pp. 52–53.

1.3 Interior norms

When interior estimates are sharply presented in terms of distance to the boundary, the natural
non-dimensionalization reveals itself. So the non-dimensionalized norms come of age. Interior
elliptic estimates are often presented in two concentric balls, in which the dimension of the
inner ball is the same as the distance to the boundary of the outer ball.

3
 Consider an open proper subset Ω ⊂ Rn . For x, y ∈ Ω, let dx = dist (x, ∂Ω), dx,y =
min(dx , dy ). We define for u ∈ C k (Ω), C k,α (Ω) the following quantities,
[u]∗k,0;Ω = [u]∗k;Ω = sup|β|=k supx∈Ω dkx |Dβ u(x)|, k = 0, 1, 2, · · · ;

|u|∗k;Ω = |u|∗k,0;Ω = kj=0 [u]∗j;Ω ;

P

β u(x)−D β u(y)|
(7)
|D
[u]∗k,α;Ω = sup|β|=k supx,y∈Ω dk+α
x,y |x−y|α , 0 < α ≤ 1;

|u|∗k,α;Ω = |u|∗k;Ω + [u]∗k,α;Ω .

In this notation, we have
[u]∗0;Ω = |u|∗0;Ω = |u|0;Ω .

The various derivatives are naturally compensated with multiplication by the distance to
the appropriate powers.

We note that the starred norms |u|∗k;Ω and |u|∗k,α;Ω are norms on the subspaces of the local
spaces C k (Ω) and C k,α (Ω) respectively for which they are finite. If Ω is bounded and d = diam
Ω, then obviously these interior norms and the global norms are related by
|u|∗k,α;Ω ≤ max(1, dk+α )|u|k,α;Ω .
If Ω0 ⊂⊂ Ω and σ = dist (Ω0 , ∂Ω), then
min(1, σ k+α )|u|k,α;Ω0 ≤ |u|∗k,α;Ω .

See [2], Sect.4.3, p. 61.

1.4 Partially interior norms (Interior with some boundary norms)

When a solution behaves smoothly on a portion of the boundary, thus the portion is not really
boundaries, the portion can be included with the interior, hence creating a set appropriately
called partially interior. Thus the distance function dx = dist (x, ∂Ω) can be restricted to the
true boundary (aka bad boundary). See Figure 1 for an image.

4
 Consider T ⊂ ∂Ω. Let

dx = dist (x, ∂Ω − T ); dx,y = min(dx , dy ).

Using this restricted distance in place of dx , we change the interior norms to “partially interior
norms” (called by [2]), which we call Interior with some boundary norms. The notations are
k
[u]∗k,0;Ω∪T = [u]∗k;Ω∪T = sup|β|=k supx∈Ω dx |Dβ u(x)|, k = 0, 1, 2, · · · ;

|u|∗k;Ω∪T = |u|∗k,0;Ω∪T = kj=0 [u]∗j;Ω∪T ;

P
(8)
k+α |Dβ u(x)−Dβ u(y)|
[u]∗k,α;Ω∪T = sup|β|=k supx,y∈Ω dx,y |x−y|α , 0 < α ≤ 1;

|u|∗k,α;Ω∪T = |u|∗k;Ω∪T + [u]∗k,α;Ω∪T .

See [2], Sect.4.3, pp. 65–66.

1.5 Intermediate norms

Interior (with some boundary) norms super-dimensionally weighted

We sometimes need more weight of the distance. For σ a real number and k a non-negative
integer we define
(σ) (σ)
[f ]k,0;Ω = [f ]k;Ω = supx∈Ω,|β|=k dk+σ
x |Dβ f (x)|;
(σ) |D β f (x)−D β f (y)|
[f ]k,α;Ω = supx,y∈Ω,|β|=k dk+α+σ
x,y |x−y|α ;
(9)
(σ) Pk (σ)
|f |k;Ω = j=0 [f ]j;Ω ;
(σ) (σ) (σ)
|f |k,α;Ω = |f |k;Ω + [f ]k,α;Ω .

In this notation, when σ = 0, these quantities are identical with those defined earlier, so that
[·](0) = [·]∗ and | · |(0) = | · |∗ .

It is easy to verify that

(σ+τ ) (σ) (τ )
|f g|0,α;Ω ≤ |f |0,α;Ω |g|0,α;Ω for σ + τ ≥ 0. (10)

When σ > 0, I call the norm super-dimensionally weighted. These norms can measure the
growth rates of the function and its derivatives at the (targeted) boundary.

Example 2: Theorem 4.8 (p.62) of [2]: Let u ∈ C 2 (Ω), f ∈ C α (Ω) satisfy ∆u = f in an

open set Ω of Rn . Then
(0) (2)
|u|2,α;Ω ≤ Cn,α (|u|0;Ω + |f |0,α;Ω ).

Interior (with some boundary) norms sub-dimensionally weighted

5
 When σ < 0, I call the norms sub-dimensionally weighted. These are the true Intermediate
norms introduced bu Gilbarg and Hörmander [1]. These norms can pick up the vanishing rates
of the function and its derivatives at the (targeted) boundary.

Example 3. Let u ∈ C 2,α (Ω) satisfy the equation Lu = f in an open set Ω ⊂ Rn ,

X X
Lu := aij (x)uxi xj + bi (x)uxi + cu (11)
X
aij ξi ξj ≥ λ|ξ|2 , ∀x ∈ Ω, ξ ∈ Rn (12)
(0) (1) (2)
|aij |0,α;Ω , |bi |0,α;Ω , |c|0,α;Ω ≤ Λ. (13)
Then there holds
(−β) (−β) (2−β)
|u|2,α;Ω ≤ C (|u|0;Ω + |f |0,α;Ω ), (14)
where C = Cn,α,β,λ,Λ for any β ∈ R (as long as the right-hand side is finite). See Lemma 6.20,
p.112 of [2].

References
[1] Gilbarg, D., and L. Hömander, Intermediate Schauder estimates, Arch. Rat. Mech. Anal.,
74(1980), 297–318.
[2] Gilbarg, D., and Trudinger, N., Elliptic Partial Differential Equations of Second Order,
Springer, New York, 2nd ed, 1983.
[3] Grisvard, P., Elliptic Problems in Nonsmooth Domains, Pitman Advanced Publishing
Program, Boston, 1985.
[4] Guan, Pengfei; Sawyer, E., Regularity estimates for the oblique derivative problem on
non-smooth domains (I), Chin. Ann. of Math. 16B: 3(1995), 299–324.
[5] Hörmander, L., Pseudo-differential operators and non-elliptic boundary problems, Ann.
of Math., 83(1966), 129–209.