2
$\begingroup$

Consider $\{0,1\}$ with the discrete topology and $\Sigma=\{0,1\}^{\mathbb{N}}$ with the product topology. We know that this product topology is generated by the metric $$ d(x,y)=\sum_{n=1}^{\infty}\dfrac{|x_i-y_i|}{2^i}. $$

Denote by $C(\Sigma)$ the space of all continuous functions $f:\Sigma \to \mathbb{R}$ and $C^{\gamma}(\Sigma)$ the subset of $C(\Sigma)$ of all $f$ such that $$ Hol(f)=\sup_{x\neq y}\dfrac{|f(x)-f(y)|}{d(x,y)^{\gamma}}<\infty $$

We consider on $C(\Sigma)$ the sup norm, i.e, $\|f\|=\sup_{x\in \Sigma}|f(x)|$ and on $C^{\gamma}(\Sigma)$ the "Hölder norm" defined by $\|f\|_{\gamma}=\|f\|+Hol(f)$.

Fact: $C^{\gamma}(\Sigma)$, $0\leq\gamma <1$ is dense in $C(\Sigma)$ in the uniform norm.

New setting: Let $(M,d)$ be a compact metric space, and $\Sigma=M^{\mathbb{N}}$ with the product topology. $\Sigma$ is compact by Tychonov's Theorem. The product topology is generated by the metric
$$ d_{\Sigma}(x,y)=\sum_{n=1}^{\infty}\dfrac{d(x_i,y_i)}{2^i} $$ In an analogously way we the spaces $C(\Sigma)$ and $C^{\gamma}(\Sigma)$ and the norms $\|\cdot\|$ and $\|\cdot\|_{\gamma}$ over $C(\Sigma)$ and $C^{\gamma}(\Sigma)$ respectively, they are both well defined because $\Sigma$ is compact.

My Question: Is, for $0<\gamma<1$, $C^{\gamma}(\Sigma)$ dense in $C(\Sigma)$ in the uniform norm?

$\endgroup$
3
  • 1
    $\begingroup$ Use Stone-Weierstrass on functions which depend only on finitely many co-ordinates? $\endgroup$
    – Ian Morris
    Commented Nov 1, 2015 at 19:13
  • 3
    $\begingroup$ In fact, the Lipschitz functions are already dense. This follows from from Stone-Weierstrass. $\endgroup$ Commented Nov 1, 2015 at 19:17
  • 3
    $\begingroup$ Also, taking the product is pointless in your general setting since: (1) you can right away take $M=\Sigma$; (2) the functions on $\Sigma$ could depend on just the first coordinate. $\endgroup$ Commented Nov 1, 2015 at 19:18

0

You must log in to answer this question.