Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

We have 3 arithmetic progressions $na_1 + c_1, na_2 + c_2, na_3 + c_3$ with the reals $a_1,a_2,a_3$ linearly independent over the rationals. Is it always true that given $\epsilon > 0$ there are terms $x_1,x_2,x_3$, one from each progression, such that all three are within $\epsilon$ of each other ?

Same problem with any finite number of progressions.

share|improve this question
    
Your question lacks motivation, please see the faq. Why are you interested in this question? What have you tried already? –  Benoît Kloeckner Nov 15 '11 at 17:50
6  
Hi Jeff, welcome to MO! –  Emil Jeřábek Nov 15 '11 at 18:01
    
I suspect that a bit weaker statement (say that this is true up to a set of some H-dim value or something of the a_{i}'s) can be derived from Lindenstrauss' (partial-)solution to the Littlewood's conjecture (see Shpaira - "ON A GENERALIZATION OF LITTLEWOOD'S CONJECTURE"). –  Asaf Nov 16 '11 at 9:24

3 Answers 3

This is actually not true. For a counterexample, take $(a_1,a_2,a_3) = (1,\phantom. \pi,\phantom.\pi/(\pi+1))$ and $(c_1,c_2,c_3) = (0,0,1)$, so our sequences are $\lbrace n_1 \rbrace$, $\lbrace n_2 \pi \rbrace$, and $\lbrace 1 + n_3 \frac\pi{\pi+1}\rbrace$. If $n_1$ is near $n_2 \pi$ then it's also near $(n_1+n_2)\frac\pi{\pi+1}$, and therefore far from $1 + n_3 \frac\pi{\pi+1}$ for any integer $n_3$.

As this example suggests, the correct condition (for any number of arithmetic progressions) is that the reciprocals of the common differences $a_i$ be linearly independent over ${\bf Q}$.

Indeed, given nonzero $a_1,\ldots,a_{k+1} \in {\bf R}$, consider the image of the lattice $L = \oplus_{i=1}^{k+1} {\bf Z} a_i \subset {\bf R}^{k+1}$ under the quotient map $\pi : {\bf R}^{k+1} \rightarrow V := {\bf R}^{k+1} / {\bf R}v$ where $v=(1,1,\ldots,1)$. Then the desired result is true for all $c_1,\ldots,c_{k+1}$ iff $\pi(L)$ is dense in $V$. But $\pi(L)$ is a subgroup of $V$, so its closure is a closed subgroup — and a closed subgroup $G \subseteq V\phantom.$ is properly contained in $V\phantom.$ iff there a nonzero linear functional $\varpi: V \rightarrow {\bf R}$ such that $\varpi(G) \subseteq \bf Z$. Hence $\pi(L)$ fails to be dense in $V\phantom.$ iff there is a nonzero functional $V \rightarrow {\bf R}$ taking $\pi(L)$ to a subset of ${\bf Z}$. But the linear functionals on $V\phantom.$ are simply the functionals on ${\bf R}^{k+1}$ vanishing on $v$; that is, the maps $(x_1,\ldots,x_{k+1}) \mapsto \sum_{i=1}^{k+1} b_i x_i$ for some $b_i\in\bf R$, not all zero, such that $\sum_{i=1}^{k+1} b_i = 0$. Hence our condition is $b_i a_i \in \bf Z$, or equivalently $b_i \in {\bf Z} a_i^{-1}$. There exist such $b_i$ that sum to zero iff the $a_i^{-1}$ are linearly dependent over the rationals, QED.

share|improve this answer
    
Good point, and when I look at my answer I see that it is the linear independence of the reciprocals I actually use, not of the actual numbers. I will update my answer to reflect this –  Johan Andersson Nov 15 '11 at 20:52

This follows from the Kronecker's approximation theorem (With the conditions modified. See edit below).

Assume without loss of generality that $a_k$ are positive, that $c_1 \leq c_k$ and $0 \leq c_k < a_k$. Let $n_1=n$ and $n_k=\lfloor na_1/a_k\rfloor$ for $k=1,\ldots,K$. Then $n_1 a_1+c_1-(n_k a_k+c_k)=n a_1 -\lfloor na_1/a_k\rfloor a_k+c_1-c_k=c_1-c_k + $ { $ n a_1/a_k $ }$a_k$

By Kronecker's approximation theorem there exists some $n$ such that $|c_1-c_k + $ {$n a_1/a_k $}$a_k|<\epsilon/2$ for each $k=1,\ldots,K$. The conclusion follows from the triangle inequality.

Edit: As Noam D. Elkies remarked in his answer what we use here is in fact the linear independence over $\mathbb Q$ of the reciprocals of the numbers, $1/a_k$ (Or equivalently in my application of the Kronecker's approximation theorem, the numbers $a_1/a_k$ ), not the numbers $a_k$ themselves. This means that the question as posed is not true, but it is true when the conditions are modified.

share|improve this answer
    
[while you're correcting: it's Kronecker, not Kroenecker (or Krönecker).] –  Noam D. Elkies Nov 15 '11 at 21:14

For given $(a_1, a_2, a_3)$ and $(c_1, c_2 , c_3)$, the required condition is equivalent to: $(c_2-c_1, c_3-c_2)$ is in the closure $\Gamma $ of the additive subgroup generated by $(a_1,0)$, $(-a_2,a_2)$, $(0,-a_3)$. But the latter need not be dense, even if $(a_1, a_2, a_3)$ are linearly independent over the rationals; for instance $(a_1, a_2, a_3):=(1+\lambda,1,1+1/\lambda)$ with $\lambda$ an irrational non quadratic, is a linear independent triple for which $\Gamma$ is a family of lines $\{(x,y)\, : \, y=\lambda x + n(\lambda+1)\, , n\in\mathbb{N}\}$.

share|improve this answer
    
Right, my example showed that with $\lambda = -(\pi+1)$. Note that $a_2^{-1} = a_1^{-1} + a_3^{-1}$, so again the reciprocals of the $a_i$ are linearly dependent over ${\bf Q}$. –  Noam D. Elkies Nov 16 '11 at 0:09
    
Yes, excellent. I realized that there where more complete answers only when I finished mine... I posted it to be in the party ;-) –  Pietro Majer Nov 16 '11 at 6:44

Your Answer

 
discard

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.