3
$\begingroup$

Let $A, B \subset \mathbb R$ be IID random closed subset. Suppose that there exists $d \in (1/2, 1]$ such that the Hausdorff dimension of $A$ is equal to $d$ almost surely. Is it true that $\mathbf P\lbrack A \cap B \neq \emptyset\rbrack > 0$?

(Note that zero sets of two independent Brownian motions almost surely do not intersect. Thus we have to require at least $d > 1/2$.)

$\endgroup$
4
  • $\begingroup$ What do you mean by IID random subset? See the Furstenberg Intersection Conjecture, for a result in this general direction. $\endgroup$ Commented Jul 16 at 17:02
  • $\begingroup$ @AnthonyQuas I mean $A$, $B$ are random subsets of $\mathbb R$, independent, and have the same law. $\endgroup$
    – Focus
    Commented Jul 16 at 17:30
  • $\begingroup$ @Focus Do you assume $A, B$ to be closed or not? I think that may make a difference. Also, could it be that the question is actually more or less "Are there uncountably many disjoint measurable (, ...) subsets of $[0, 1]$ such that everyone of them has Hausdorff dimension $d > 1/2$"? If this is the case, maybe set theorists/logicians know this. $\endgroup$
    – unwissen
    Commented Jul 17 at 10:12
  • $\begingroup$ @unwissen Yes, we may assume that they are closed, thanks. $\endgroup$
    – Focus
    Commented Jul 17 at 11:49

1 Answer 1

3
$\begingroup$

No. They may be almost surely disjoint.

If $y\in\{0,1\}^{\mathbb N}$, let $S_y$ denote the (closure of the) set of $x$'s in $[0,1]$ such that the $2^n$th binary digit of $x$ is $y_n$; and the $(2^n+1)$st and $(2^n-1)st$ digits of $x$ are 0. [These additional conditions are just to prevent issues with multiple binary expansions].

If $y$ is selected uniformly from a coin-tossing measure, then $S_y$ and $S_{y'}$ are almost surely disjoint. And the Hausdorff dimension of each $S_y$ is 1: The box dimension is obviously 1: the number of balls of diameter $2^{-N}$ to cover $S_y$ is $2^{N-3\log_2 N}$; obtaining the Hausdorff dimension requires Frostman's lemma.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .