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

Let $\alpha>0$ and define $S(\alpha)=\{\lfloor n \alpha \rfloor: n\in\Bbb Z^+ \}$. (Here $\lfloor x\rfloor$ is the integer part of $x$ and $\mathbb Z^+$ the set of positive integers.)

Is there any known characterization for the pairs of positive irrationals $\alpha$ and $\beta$ for which $S(\alpha)\cap S(\beta)=\varnothing$?

share|cite|improve this question
See also… – Gerry Myerson Dec 11 '13 at 22:19

Not a complete answer: these sequences are called Beatty sequences. It's a nice exercise to show that $S(\alpha)$ and $S(\beta)$ partition the positive integers when $\frac{1}{\alpha} + \frac{1}{\beta} = 1$, hence for $\alpha, \beta$ satisfying this condition we can take $S(n \alpha), S(m \beta)$ where $n, m \in \mathbb{N}$.

share|cite|improve this answer
It seems to me that $S(\alpha)\cap S(\beta)=\varnothing$ if and only if $m/\alpha+n/\beta=1$, for some $m,n$ positive integers. – smyrlis Dec 11 '13 at 20:59

You are right about $S(\alpha)\cap S(\beta)=\varnothing\iff \dfrac{n}{\alpha}+\dfrac{m}{\beta}=1$ for some $n,m\in\mathbb Z^+$ (see Theorem 8 of the cited paper).

The implication $\Longleftarrow$ is easy so let's assume that $\frac{n}{\alpha}+\frac{m}{\beta}\neq 1$ for all $n,m\in\mathbb Z^+$. Then one of the following is true.

  1. The numbers $1,\dfrac{1}{\alpha},\dfrac{1}{\beta}$ are linearly independent over $\mathbb Q$,
  2. there exist some $n,m,k\in\mathbb Z^+$ such that $\left|\dfrac{n}{\alpha}-\dfrac{m}{\beta}\right|=k$,
  3. there exist some $n,m,k\in\mathbb Z^+$ such that $\dfrac{n}{\alpha}+\dfrac{m}{\beta}=k$ with $k>1$, and $\gcd(n,m,k)=1$.

It is enough to show that any one of 1., 2. or 3. implies that $S(\alpha)\cap S(\beta)=\emptyset$. Elementary proofs of these can be found in this paper (Theorem 5, Theorem 6 and Theorem 7 respectively).

share|cite|improve this answer

I suggest looking in Joe Roberts amazing book I thought I remembered a partition into 3 such sequences but actually it is $\lfloor \tau \lfloor \tau n \rfloor \rfloor$,$\lfloor \tau \lfloor \tau^2 n \rfloor \rfloor$,$\lfloor \tau^2 n \rfloor $ So not exactly what was requested. Maybe that is not so deep.

I put in this answer partly to get the link to the book in.

share|cite|improve this answer
It was a problem on the 1995 Putnam exam that there is no partition of the natural numbers into three sequences $\lfloor n\alpha\rfloor$, $\lfloor n\beta \rfloor$, and $\lfloor n\gamma\rfloor$. – Lucia Dec 12 '13 at 1:21
Indeed, it is possible to prove that three such sequences are never pairwise disjoint. – Rodrigo Dec 12 '13 at 2:29
Ha. OP has posted the 3-sequence question to… – Gerry Myerson Dec 12 '13 at 5:23
The Putnam exam version has easier proof, because the statement is weaker. – i707107 Dec 13 '13 at 18:59

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.