2
$\begingroup$

Fix $ k \in \mathbb{N} $ and let $ H^k $ be the $k$-dimensional Hausdorff measure on $\ell^\infty $. Also, if $ V $ is a subspace of $ \ell^\infty $, we denote the projection onto $ V $ by $ \pi_V $. Let $ A $ be a subset of $ \ell^\infty $ such that $ H^k(\pi_V(A)) = 0 $ for all finite dimensional subspaces $ V $ of $ \ell^\infty $.

Can we conclude that $ H^k(A) = 0 $?

Any help would be appreciated.

Thanks.

$\endgroup$
  • 2
    $\begingroup$ For $\ell^2$ the answer seems to be no, for orthogonal projections. For $\ell^\infty$ it is not clear which metric you want to consider on $A$ and which projections $\endgroup$ – Anton Petrunin Nov 10 '13 at 8:42
  • $\begingroup$ The standard metric on $ \ell^\infty $ which is the supremum metric. And you're right. I should've been more careful about the projections. I meant to say projection onto the first m components. $ \pi_V(x_1, x_2, \dots) = (x_1, \dots, x_m) $. $\endgroup$ – Axiom Nov 10 '13 at 17:41
  • 1
    $\begingroup$ In this case I think the answer should be no (see my answer below). Maybe assuming the stronger condition $H^k(\pi(A))=0$ for any projector $\pi$ with finite dimensional range allows a positive answer? $\endgroup$ – Pietro Majer Nov 10 '13 at 21:24
4
$\begingroup$

For any $m\in\mathbb{N}$ let $\pi_m$ denote the projector on the space $V_m:=\{x\in \ell_\infty\, : \, x_j=0,\,\, \forall j>m\}$. Let $S$ denote the left shift operator on $ \ell_\infty$, that is $Sx:=( x_2,x_3,\dots)$ for any $x:=(x_1,x_2\dots)\in \ell_\infty$. Consider the bounded linear operator (a norm-$2$ projector onto $V_m$, indeed)

$$P_m:=\pi_m(I-S^m/2)^{-1}=\pi_m \sum_{j=0}^\infty 2^{-j}S^{jm}\, .$$ Finally, consider the set $A:=\{0,1\}^\mathbb{N_+}=\{x\in \ell_\infty\, : \forall j\,\, x_j\in\{0,1\} \} $. For any $m$ the set $\pi_m(A)=\{0,1\}^m$ is finite, so $\mathcal{H}^k(\pi_m(A))=0$ for any $k>0$. On the other hand, $P_m(A)=[0,2]^m$, so $2^m=\mathcal{H}^m(P_m A)\le \|P_m\|^m \mathcal{H}^m(A)$ for any $m$, whence $\mathcal{H}^m(A)\ge 1$ for any $m$, so in fact $\mathcal{H}^m(A)= \infty$ for any $m$.

$\endgroup$

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.