For $A\subseteq\mathbb{N}$ we define the upper density to be $$\text{ud}(A) = \lim\sup_{n\to\infty}\frac{|A\cap\{1,\ldots,n\}|}{n}$$

Is there an infinite set ${\cal S}$ of pairwise disjoint subsets of $\mathbb{N}$ such that $\text{ud}(S) > 0$ for all $S\in {\cal S}$?


Yes. Take the segments of integers $B_i=\{2^i,\dots,2^{i+1}-1\}$. Partition naturals onto disjoint infinite sets $I_1,\dots$. Take the sets $A_i=\cup_{k\in I_i} B_k$.

  • $\begingroup$ You should emphasize that this realizes not only a positive upper density, but upper density $1$ for each of the $A_i$, thus answering a stronger and more interesting question than OP asked. $\endgroup$ – Gro-Tsen Sep 6 '17 at 9:40
  • $\begingroup$ Oh yes, just take $B_i=\{i!+1,\dots,(i+1)!\}$. $\endgroup$ – Fedor Petrov Sep 6 '17 at 10:07

Yes there is. For $n\in\mathbb N$ take $S_n=\{k\in\mathbb N|k\equiv 2^{n-1}\pmod {2^n}\}$.
That is $S_1$ are the odd numbers, $S_2$ their doubles etc. Obviously $\bigcup S_n=\mathbb N$ and $ud(S_n)=2^{-n}$.

  • 1
    $\begingroup$ You can extend this idea by taking any discrete valuation $\nu_p$ on the integers for a prime $p$ and construct the sets $S_n = \{k \in \mathbb{N} \mid \nu_p(k) = n\}.$ This should also generalize to many other rings. $\endgroup$ – Dirk Sep 6 '17 at 9:01

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.