0
$\begingroup$

A simple question from someone new to the field:

In a metric space, the Hausdorff dimension of a subset is defined by covering the subset with $\epsilon$-balls and looking at how the number of required balls grows as a power or $\epsilon$ in the limit $\epsilon \to 0$.

My question is this: If my space is $\mathbb{R}^n$ and the metric $d_p$ is the metric induced by the $p$-norm, $p \in (1, \infty]$, does the Hausdorff dimension of a subset $A \subseteq \mathbb{R}^n$ depend on the choice of $p$?

(My initial thought was that every $d_p$-ball has the same $d_q$-dimension for all $q$ (namely $n$), so that I believe all the dimensions should coincide, but I'm sure I'm overlooking something.)

Thanks!

$\endgroup$
  • 6
    $\begingroup$ Hausdorff dimension is a bi-Lipschitz invariant, and all norms of $\mathbb{r}^n$ are equivalent, so the Hausdorff dimension does not depend upon the norm. $\endgroup$ – Benoît Kloeckner Nov 3 '10 at 11:30
3
$\begingroup$

Let $B_p$ denote the 1-ball with centre 0 with respect to the $l^p$ norm. For any $p$ and $q$ there is a number $N$ such that $B_p$ is covered by $N$ translates of $B_q$. Then any $\epsilon$-ball in the $l^p$ norm is covered by $N$ $\epsilon$-balls in the $l^q$ norm. Thus within a constant factor, the number of $\epsilon$-balls required to cover a set in the $l^p$ and $l^q$ norms is the same. This constant factor won't affect the asymptotic power in the number of $\epsilon$-balls required to cover a given set, so the Hausdorff dimension in both cases is the same.

$\endgroup$
  • $\begingroup$ It is better to say that $(A,\ell^p)$ is bi-Lipschitz equivalent to $(A,\ell^q)$ and therefore their Hausdorff dimensions coincide. $\endgroup$ – Anton Petrunin Nov 3 '10 at 18:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.