MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

If {c(n)} is an arbitrary sequence of irrational numbers converging to 0 then Q + c(n), the set obtained by adding c(n) to the set of rational numbers Q, is clearly disjoint from Q for each n.

Is there an uncountable dense set of the real numbers, say D, for which a sequence {c(n)} converging to 0 exists such that D + c(n) does not intersect D for all n?

If such a set exists, and is borel, then it must have measure 0 since D - D would contain an interval if D had a positive measure.

share|cite|improve this question
Does anyone know if the following problem (a special version of Erdos similarity problem) is still open: Is there a measurable set of reals with positive measure that doesn't contain a similar copy of the sequence (1/2)^n? – Ashutosh Jul 20 '10 at 19:32
up vote 3 down vote accepted

Choose your favorite c(n), and define $D$ as any complementary $\mathbb{Q}$-vector space of $\mathrm{span}(\{c(n): n\in\mathbb{N}\})$ in $\mathbb{R}$ as $\mathbb{Q}$ vector space. This $D$ is not measurable, of course (countably many translates of it cover $\mathbb{R}$, so it can't be of measure zero, and on the other hand as you said it can't be of positive measure). It's a version of the construction of the Vitali set (actually the standard Vitali set already works with a sequence of rationals c(n) ).

share|cite|improve this answer
I am sorry, I didn't get you, could you expand a bit? – acarchau Jul 19 '10 at 21:29
I understood it Daniel's comment, thanks. – acarchau Jul 19 '10 at 21:32
expanded enough? – Pietro Majer Jul 19 '10 at 21:36
Yes. This is quite clear. – acarchau Jul 19 '10 at 21:39
note: the existence of a complementary space to $\mathbb{Q}$ requires the axiom of choice, but as François G. Dorais remarks, you just need an uncountable rational subspace with$ D\cap \mathbb{Q}=(0)$, that can be exhibited explicitly. – Pietro Majer Jul 21 '10 at 7:49

Use of the Axiom of Choice is unnecessary here. One can write down an explicit $\mathbb{Q}$-linearly independent subset $\mathcal{T}$ of $\mathbb{R}$ with size $2^{\aleph_0}$ as I wrote in this answer. If it so happens that some of the $c(n)$ are in the $\mathbb{Q}$-span of this set $\mathcal{T}$, then remove the finitely or countably many elements which are involved in expressing the $c(n)$ as a $\mathbb{Q}$-linear combination of elements of the original set $\mathcal{T}$; what remains of $\mathcal{T}$ still has size $2^{\aleph_0}$. The $\mathbb{Q}$-span of this remainder has the desired property. The set in question will be a Borel set (of Lebesgue measure 0).

share|cite|improve this answer

Yes. We proceed using the axiom of choice. Choose an basis of $\mathbb{R}$ with cardinality $|\mathbb{R}/\mathbb{Q}|$ and a bijection between this basis and the elements of $\mathbb{R}/\mathbb{Q}$. Pick an element from each coset of $\mathbb{Q}$ contained in the corresponding basis set (we may do this as each coset is dense); this will be our set $D$. It is dense as each basis set has non-empty intersection with $D$. Now any sequence of rationals tending to zero works as $c(n)$, as if $D+c(n)$ had non-empty intersection with $D$, then two elements of $D$ would differ by a rational.

share|cite|improve this answer
Ah, I see Pietro's answer is identical to mine (and preceded mine chronologically); note that it also requires the axiom of choice. – Daniel Litt Jul 19 '10 at 21:29
Why should D be dense? – acarchau Jul 20 '10 at 15:15
You're right, I missed that. Should be fixed now. – Daniel Litt Jul 20 '10 at 15:47

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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