1

I am trying to solve the following problem:

Find a set $X \subset \Bbb R$ s t.$\dim_H (X)= s$ where $s = \frac{\log 2}{\log 3}$, but $H^s (X) = \infty$.

Here I am using the notations from Fractal Geometry by Kenneth Falconer. From exercise 4.9 of the book. I think I should be looking at some sort of intersection of middle $\lambda$-Cantor sets with $\lambda$ getting closer to $\frac13$ from below. However, I am not sure where to go from there.

Edit: after everyone's suggestions here is what I came up with. Let $C_n$ be a rescale of the middle third Cantor set (call it $C$) which has been translated to fit in $[ \frac{1}{2n}, \frac{1}{2n-1}]$. Let $X = \cup_{n=1}^{\infty} C_n$.

Note $C_n$ and $C_{n+1}$ are $\frac{1}{2n} - \frac{1}{2n+1} = \frac{1}{(2n+1)2n}$ apart. Fix $\varepsilon > 0$ and find $N = N_{\varepsilon}$ such that \begin{equation} \frac{1}{(2N+3)(2N+2)} \leq \varepsilon < \frac{1}{(2N+1)2N}. \end{equation} It follows that \begin{equation} H^s_{\varepsilon} (X) \geq \sum_{n=1}^{N_{\varepsilon}} H^s_{\varepsilon} (C_n) = \sum_{n=1}^{N_{\varepsilon}} H^s_{2n(2n-1)\varepsilon} (C). \end{equation} The rest is clever algebra to show that the above sum diverges to $\infty$ using the fact that $H^s (C) > 0$.

Wolfgangg
  • 189
  • Did you mean $\log\color{red}2/\log3$ in the title? – J. W. Tanner May 22 '20 at 16:55
  • Yes, I want a set with Hausdorff dimension equal to $\frac{\log 2}{\log 3}$ – Wolfgangg May 22 '20 at 17:06
  • To mention some user, write @username. – PinkyWay May 22 '20 at 17:07
  • I like your idea of taking a union of sets. I'd recommend taking a union of scaled Cantor ternary sets, though. – Mark McClure May 22 '20 at 19:53
  • 1
    Or you could translate instead of scaling. – GEdgar May 22 '20 at 20:04
  • In a set is $s$-dimensional (in the sense of the Hausdorff dimension), then it could have infinite, zero, or finite-but-nonzero $s$-dimensional Hausdorff measure. A natural question is to consider how to make each of these three things happen. However, before you start trying to tackle something "pathological", why not try to come up with examples in more familiar spaces. For example, can you think of a $1$-dimensional set with infinite $1$-dimensional Hausdorff measure? – Xander Henderson May 23 '20 at 05:12
  • @GEdgar Thank you for your suggestion. I have updated my solution. Would you be kind enough to have a quick read and tell me if this looks right? – Wolfgangg May 23 '20 at 09:25

0 Answers0