3
$\begingroup$

Let $\mathbb Z^+$ be the set of positive integers. In 1934, Romanoff proved that $$\liminf_{x\to+\infty}\frac{|\{n\le x:\ 2n+1=p+2^k\ \text{for some prime}\ p\ \text{and}\ k\in\mathbb Z^+\}|}x>0.$$ On the other hand, in 1950 van der Corput showed that $$\liminf_{x\to+\infty}\frac{|\{n\le x:\ 2n+1\not=p+2^k\ \text{for any prime}\ p\ \text{and}\ k\in\mathbb Z^+\}|}x>0.$$ These two results are well-known, and they have stimulated some further work.

For $n\in\mathbb Z^+$, the central binomial coefficient $\binom{2n}n=2\binom{2n-1}{n-1}$ is even. By Stirling's formula $n!\sim\sqrt{2\pi n}(n/e)^n$, we have $$\binom{2n}n=\frac{(2n)!}{(n!)^2}\sim \frac{4^n}{\sqrt{n\pi}}.$$

Motivated by the results of Romanoff and van der Corput, here I ask the following question.

QUESTION. Let $$S=\left\{p+\binom{2k}k:\ p\ \text{is an odd prime and}\ k\in\mathbb Z^+\right\},$$ and define $$s_1=\liminf_{x\to+\infty}\frac{|\{n\le x:\ 2n+1\in S\}|}x \ \ \text{and}\ \ s_2=\liminf_{x\to+\infty}\frac{|\{n\le x:\ 2n+1\not\in S\}|}x.$$ Is it true that both $s_1$ and $s_2$ are positive? Can one modify Romanoff's method or van der Corput's method to prove $s_1>0$ or $s_2>0$.

By the Prime Number Theorem, $\pi(x)\sim x/(\log x)$. Note also that $1/\log4>1/2$. Based on this and my computation, I conjecture that the two constants $s_1$ and $s_2$ are indeed positive.

Any comments are welcome!

$\endgroup$

1 Answer 1

8
$\begingroup$

The limit (not only liminf) $s_1$ equals 0, i. e., $|S\cap [1,x]|=o(x)$.

The reason is that the central binomials are almost always divisible by 3, 5 etc (and by every specific prime). To be more precise, we have

Observation. for every fixed $p$ we have $\bigl|n\leqslant x: p\nmid{2n\choose n}\bigr|=o(x)$.

Proof. Recall that by Kummer's theorem $p$ does not divide ${n+k\choose k}$ if and only if the base $p$ expansions $n=\sum c_i p^i$ and $k=\sum b_i p^i$, $0\leqslant b_i,c_i\leqslant p-1$, satisfy $b_i+c_i\leqslant p-1$ for all $i$ (i.e., there are no carries if we sum up $n$ and $k$ in base $p$). For ${2n\choose n}$ this means that all base $p$ digits of $n$ must be less than $p/2$, there are at most $\lceil p/2\rceil^{1+\log_p x}=o(x)$ such numbers not exceeding $x$.

Thus, for every fixed $M$ the number of $n\leqslant x$ for which there exists prime $p<M$ not dividing ${2n\choose n}$ is $o(x)$. Fix $M$ and call such $n$ $M$-strange.

Counting numbers of the form $p+{2n\choose n}$ not exceeding $x$, we note that there exist $O(x/\log x)$ choices of $p$ and $o(\log x)$ choices of $M$-strange $n$. Totally this gives $o(x)$ numbers of the form $p+{2n\choose n}$ with $M$-strange $n$. For not $M$-strange $n$, the number $y=p+{2n\choose n}\leqslant x$ either satisfies $p\leqslant M$ (there are $O(\log x)$ such numbers), or $y$ does not have a prime divisor $q<M$, since $q$ divides ${2n\choose n}$ and does not divide $p$ (there are at most $x\cdot \prod_{p<M}(1-1/p)+o(x)$ such numbers $y$). Since the product $\prod_p (1-1/p)$ equals 0, we get that there are only $o(x)$ numbers less than $x$ of the form $p+{2n\choose n}$.

$\endgroup$
0

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.