logo

Holder Continuous Function Spaces 📂Hilbert Space

Holder Continuous Function Spaces

Definitions1

Space of Continuous Functions

Let ΩRn\Omega \subset \mathbb{R}^{n} be called an open set. For a non-negative integer mm, for all multi-indices α\alpha where αm|\alpha| \le m, the set of ϕ\phi that are continuous in Ω\Omega for DαϕD^{\alpha}\phi is called the space of continuous functions.

Cm(Ω):={ϕ:Dαϕ is continuous on Ω,α<m} C^{m}\left( \Omega \right) := \left\{ \phi : D^{\alpha} \phi \text{ is continuous on } \Omega, \forall \left| \alpha \right| \lt m \right\}

In particular, C0(Ω):=C(Ω)C^0(\Omega):=C(\Omega), C:=m=0Cm(Ω)C^{\infty}:=\cap_{m=0}^{\infty}C^m(\Omega) is defined. Also, here DαϕD^{\alpha} \phi can also mean the weak derivative (distributional derivative) of ϕ\phi.

Space of Bounded Continuous Functions

For all 0αm0 \le |\alpha| \le m, the set of ϕCm(Ω)\phi \in C^m(\Omega) that are bounded on Ω\Omega by DαϕD^\alpha \phi is defined as the space of bounded, continuous functions.

CBm(Ω):={ϕCm(Ω):Dαϕ is bounded on Ω,α<m} C^{m}_{B}\left( \Omega \right) := \left\{ \phi \in C^{m}(\Omega) : D^{\alpha} \phi \text{ is bounded on } \Omega, \forall \left| \alpha \right| \lt m \right\}

Then naturally, CBm(Ω)Cm(Ω)C^{m}_{B}\left( \Omega \right) \subset C^{m}\left( \Omega \right) holds by definition. Furthermore, CBm(Ω)C^m_{B}(\Omega) is a Banach space given with the following norm.

ϕ ; CBm(Ω):=max0αmsupx ΩDαϕ(x) \left\| \phi\ ;\ C^m_{B}( \Omega )\right\| := \max \limits_{0 \le \left| \alpha \right| \le m } \sup \limits_{x\ \in \Omega} | D^{\alpha}\phi (x) |

Space of Bounded, Uniformly Continuous Functions

For all 0αm0 \le |\alpha| \le m, the set of ϕCm(Ω)\phi \in C^{m}(\Omega) that are bounded and uniformly continuous on Ω\Omega is called the space of bounded, uniformly continuous functions.

Cm(Ω):={ϕCm(Ω):Dαϕ is bounded and uniformly continuous on Ω,α<m} C^{m}\left( \overline{\Omega} \right) := \left\{ \phi \in C^{m}(\Omega) : D^{\alpha} \phi \text{ is bounded and uniformly continuous on } \Omega, \forall \left| \alpha \right| \lt m \right\}

Then, Cm(Ω)C^{m}\left( \overline{\Omega} \right) is a closed subspace of CBm(Ω)C^{m}_{B}\left( \Omega \right), and is a Banach space given with the following norm.

ϕ ; Cm(Ω):=max0αmsupx ΩDαϕ(x) \left\| \phi\ ;\ C^m(\overline{\Omega} )\right\| := \max \limits_{0 \le |\alpha | m } \sup \limits_{x\ \in \Omega} | D^{\alpha}\phi (x) |

Space of Hölder Continuous Functions

Hölder Condition

Let’s say 0λ10 \le \lambda \le 1. For all αm\left| \alpha \right| \le m, DαϕD^{\alpha} \phi, if there exists a constant KK satisfying the equation below, then DαϕD^{\alpha} \phi satisfies the Hölder condition on Ω\Omega.

Dαϕ(x)Dαϕ(y)Kxyλ, x,yΩ \left| D^{\alpha} \phi (x) - D^{\alpha}\phi (y) \right| \le K |x-y|^\lambda,\quad \forall\ x,y \in \Omega


The set of ϕ\phi among the elements of Cm(Ω)C^{m}\left( \overline{\Omega} \right) that satisfy the Hölder condition is called the spaces of Hölder continuous functions.

Cm,λ(Ω):={ϕCm(Ω):Dαϕ satisfies in Ω a Hoelder condition of exponent λ,αm} C^{m,\lambda}\left( \overline{\Omega} \right) := \left\{ \phi \in C^{m}\left( \overline{\Omega} \right) : D^{\alpha}\phi \text{ satisfies in } \Omega \text{ a Hoelder condition of exponent } \lambda, \forall \left| \alpha \right| \le m \right\}

Then, Cm,λ(Ω)C^{m,\lambda}\left( \overline{\Omega} \right) becomes a Banach space given with the following norm.

ϕ ; Cm,λ(Ω):=ϕ ; Cm(Ω)+max0αmsupx,y ΩxyDαϕ(x)Dαϕ(y)xyλ \left\| \phi\ ;\ C^{m,\lambda} (\overline{\Omega} )\right\| := \left\| \phi\ ;\ C^{m} (\overline{\Omega} )\right\| + \max \limits_{0 \le |\alpha | m } \sup \limits_{\substack{x,y\ \in \Omega \\ x\ne y} } \dfrac{ | D^{\alpha}\phi (x) - D^{\alpha}\phi (y) |}{ |x-y| ^{\lambda}}

The following relationship holds for 0<ν<λ10 \lt \nu \lt \lambda \le 1.

Cm,λ(Ω)Cm,ν(Ω)Cm(Ω) C^{m,\lambda}\left( \overline{\Omega} \right) \subsetneq C^{m,\nu}\left( \overline{\Omega} \right) \subsetneq C^{m}\left( \overline{\Omega} \right)

Explanation

The Lipschitz condition can be considered as the Hölder condition when λ=1\lambda = 1.


  1. Robert A. Adams and John J. F. Foutnier, Sobolev Space (2nd Edition, 2003), p10-11 ↩︎