I suppose this is une cause perdue, but it would be nice if the following held.

Let $\theta$ be an irrational, and let $c_m = {{2m}\choose m}$, the central binomial. For a real number $r$, let $d(r)$ denote the distance from $r$ to the nearest integer (often indicated by $\| r \|$). Is it true that for every irrational $\theta$, the Cesaro sums behave reasonably well: $$ (*) \quad \limsup_{N\to \infty} \frac 1N \sum_{m=1}^{N} d(c_m \theta) > 0 \,? $$ Equivalently, does there exist $c >0$---depending on $\theta$---such that $$ \limsup_{N\to \infty} \frac{\left| \left\{m \leq N\left.\right| d(c_m \theta) > c\right\}\right|}N > 0\,? $$

For almost all $\theta$, $d(c_m\theta)$ is uniformly distributed, but not for all irrational $\theta$ (since the ratio of consecutive coefficients exceeds $1$). I don't know much more about it. It's still possible (although seemingly remote) that (*) holds for all irrational $\theta$. We can also replace the central binomial terms by the Catalan numbers and the questions are probably equivalent. (An affirmative answer to either would be fine.)

A weaker version would be whether (for all irrational $\theta$) $$ \limsup_{N\to \infty} \frac{\left| \left\{m \leq N\left.\right| d(c_m \theta) > c\right\}\right|}{N^{1/2 + \delta}} = \infty $$ for some $\delta > 0$ (depending on $\theta$); an affirmative answer would be useful.

  • $\begingroup$ May help the question if you wrote it in a way that it is immediate at first look what is known and what is to be proved -- just my "even prime number" cents :-) $\endgroup$ – Suvrit Aug 10 '15 at 14:42

I don't have an answer, but I can show that it's not true if you replace $\limsup$ by $\liminf$.

Let $F_N(\theta) = \dfrac{1}{N} \sum_{m=1}^N d(c_m \theta)$ for $\theta \in \mathbb R$. These are continuous functions of $\theta$.

If $\theta$ is rational $a/\prod_{j=1}^k p_j^{d_j}$, $d(c_m \theta) = 0$ when the $p_j$-adic orders $\nu_{p_j}(c_m) \ge d_j$ for all $j$, and using Kummer's theorem it's easy to prove that the fraction of $m \in \{1,2,\ldots,N\}$ for which this is true goes to $1$ as $N \to \infty$. Thus $F_N(\theta) \to 0$ as $N \to \infty$ for rational $\theta$.

As a consequence of this and the fact that $\{\theta: F_N(\theta) < t\}$ is open, $\{\theta: \liminf_{N \to \infty} F_N(\theta) = 0\}$ is a dense $G_\delta$. On the other hand, since the mean of $F_N(\theta)$ over any interval goes to $1/4$ as $N \to \infty$, $\{\theta: \limsup_{N \to \infty} F_N(\theta) \ge 1/4 \}$ is also a dense $G_\delta$. So a generic $\theta$ (in the Baire category sense) has $\liminf_{N \to \infty} F_N(\theta) = 0$ and $\limsup_{N \to \infty} F_N(\theta) \ge 1/4$.

| cite | improve this answer | |

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.