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

I'm trying to prove (or disprove) the following "conjecture".
Given the following set of powers of two:

$$A = \{ x \mid x = 2^n \text{ and } 2^{n-1} < 3^m < x < 3^{m+1} < 2^{n+1}\}$$

(informally $2^n$ is "snapped" in an interval $[3^m,3^{m+1})$ that contains only one power of two)

Q1: Given an integer $k \geq 1$, is there a way to pick an infinite subset of $B \subseteq A$ (or possibly the whole $A$) such that the following holds:

For all integers $x,y,p,q \geq 1$, with $x,y \in B$, $x \neq y$ and all the factors of $p, q$ are smaller than $k$, does there always exist $n \geq 1$ such that:

$$x + np \in A,\quad y + nq \notin A\quad \text{?}$$

Informally whatever pair of elements of $B$ and deltas $p,q$ I choose, the two arithmetic progressions $x' = x + np$ and $y' = y +nq$ don't guarantee that $x', y'$ are always both inside or both outside of $A$.

Q2: And is the "conjecture" easier if the set is: $$A' = \{ x \mid 2^{n-1} < 3^m \leq x < 2^n < 3^{m+1} < 2^{n+1}$$

The only fuzzy idea for Q2 I have is that if $x,y$ are big enough and belong to the same $3^m \leq x<y < 2^n$ interval, then $p,q$ (that have factors smaller than $k$) don't give enough "resolution" (??) to keep $x'$ and $y'$ "aligned".

Sorry for the possibly inaccurate terminology but I'm not an expert.

share|cite|improve this question
I suspect that irrationality of log3/log2 will yield that any subset B will work, and that Q2 is not essentially different from Q1 in this regard. You might try showing that having two such a.p.'s agreeing on A would imply the ratio above is rational, although I don't know yet how to do that. – The Masked Avenger Mar 11 '14 at 16:17
@TheMaskedAvenger: I agree, I think that it's enough to pick B=A ... but I have no idea on what tools are needed to prove it, too (and my math knowledges are poor :( ... perhaps something like a very basic version of Dirichlet's theorem (??) and picking $n = n'x$, $x' = x(1 + n'p)$, $y' = y + n'xq$ – Marzio De Biasi Mar 11 '14 at 16:42
First show that as an indicator function, np+x is nonperiodic for any x. That should get you the case q=p. If q is different from p, what would it mean if the two ap's did agree? – The Masked Avenger Mar 11 '14 at 16:55
@TheMaskedAvenger: I added another condition that can simplify the problem (the factors of $p, q$ can be made arbitrarily smaller than $x$ and $y$ ...) in other words $y$ and $xq$ or $x$ and $yp$ can be considered coprimes ... but I don't know if it can help. – Marzio De Biasi Mar 11 '14 at 17:07
The members of $A$ are tabulated at – Gerry Myerson Mar 11 '14 at 22:49

Your Answer


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

Browse other questions tagged or ask your own question.