Multiplicative largeness of de Polignac numbers

Abstract

A number $m$ is said to be a de Polignac number, if infinitely many pairs of consecutive primes exist, such that $m$ can be written as the difference of those consecutive prime numbers. Recently in [B24], using arguments from the Ramsey theory, W. D. Banks proved that the collection of de Polignac number is an $IP^{\star}$ set. As a consequence, we have this collection as an additively syndetic set. In this article, we show that this collection is also a multiplicative syndetic set. In our proof, we use combinatorial arguments and the tools from the algebra of the Stone-Čech compactification of discrete semigroups (for details see [HS12]).

Keywords: Difference set of Primes, Twin prime conjecture, de Polignac numbers, Ramsey theory, $IP$-set,$IP_{r}$-set, piecewise syndetic set, algebra of the Stone-Čech compactification
Mathematics subject classification: Primary 37A44, 05D10; Secondary 11E25, 11T30.

1 Introduction

In this article, we study the Ramsey theoretic behavior of the set of de Polignac numbers (numbers that can be written as a difference between two consecutive primes in infinitely many ways), $\mathbf{POL}$ in short. This set is directly related to the twin prime conjecture. In [P16], Pintz proved that the difference set of primes has a bounded gap. Later using Maynard [M15]-Tao theorem, W. Huang and X. Wu [HW17] improved this result and proved that the difference set of primes is much larger. In fact, they proved that this set has a bounded gap in both the additive and multiplicative senses. Some recent development in this direction has been done in [G23, GHW24]. In this article we consider a much more thin subset of the difference of primes. We let $\mathbf{POL}$ be the set of all numbers which can be written as the difference of consecutive primes in infinitely many ways. In a recent work [B24], W. D. Banks proved that the set $\mathbf{POL}$ is “enough large” in $(\mathbb{N},+)$. He used Banks-Freiberg-Turnage-Butterbaugh Theorem [BFT15] and Ramsey’s theorem [R29]. However, a simple application of the theory of ultrafilters shows that the Banks theorem [B24] immediately implies that the set $\mathbf{POL}$ has additively bounded gaps (we postpone it till the end of this section). In this article we prove that the set $\mathbf{POL}$ has also multiplicatively bounded gaps.

1.1 Ramsey theoretic large sets

The notion of largeness is intimately related to the Ramsey theory. For the detailed properties of these sets, we refer the readers to [F81, HS12]. In [F81], a relation with Topological dynamics, and in [HS12] a relation with the theory of ultrafilters can be found. Let $(S,\cdot)$ be any discrete semigroup. For any $s\in S$, and $A\subseteq S$, define $s^{-1}A=\{t:s\cdot t\in A\}$. Let us recall the following notions of largeness.

Definition 1.1.

[HS12] If $(S,\cdot)$ be any discrete semigroup, then $A\subseteq S$ is said to be

1. syndetic if there exists a finite set $F\subset S$ such that $S=\cup_{t\in F}t^{-1}A.$

2. thick if for any finite set $F\subset S,$ there exists $x\in S$ such that $F\cdot x=\{y\cdot x:y\in F\}\subset A.$

3. piecewise syndetic if $A$ can be written as the intersection of syndetic and thick sets. An equivallent formulation says that $A$ is piecewise sydetic if there exists a finite set $F$ such that $\cup_{t\in F}t^{-1}A$ thick.

4. an $IP$ set if there exists an infinite sequence $\langle x_{n}\rangle_{n}$ in $S$ such that $A=FP(\langle x_{n}\rangle_{n})=\{x_{i_{1}}x_{i_{2}}\cdots x_{i_{k}}:\{i_{1}<i_{2}<\cdots i_{k}\}\subset\mathbb{N}\}$.

5. an an $IP_{r}$ set for some $r\in\mathbb{N},$ if there exists a sequence $\langle x_{n}\rangle_{n=1}^{r}$ in $S$ such that $A=FP(\langle x_{n}\rangle_{n})=\{x_{i_{1}}x_{i_{2}}\cdots x_{i_{k}}:\{i_{1}<i_{2}<\cdots i_{k}\}\subset\{1,2,\ldots,r\}\}$.

6. an $IP^{\star}$ (respectively $IP_{r}^{\star}$) if and only if $A$ intersects every $IP$ sets (resp. $IP_{r}$ sets).

Throughout our work, we will be concerned with additive and multiplicative structures of $\mathbb{N}.$ Hence while we mention a set is additively (resp. multiplicatively) large, then we mean that the set is large in $(\mathbb{N},+)$ (resp. $(\mathbb{N},\cdot)$).

For any two set $A,B\subseteq\mathbb{N},$ define $A-B=\{a-b:a\in A,b\in B,a>b\}$. Letting $\mathbb{P}$ be the set of primes, $\mathbb{P}-\mathbb{P}$ be the set of differences of primes. Recently in [B24], W. D. Banks proved the following theorem.

Theorem 1.2 (Banks theorem).

If $A\subseteq 2\mathbb{N}$ is any $IP$ set, then $\mathbf{POL}\cap A\neq\emptyset.$

In the later subsection, we show that this is true for any $IP$ set. However, the main purpose of this paper is to prove the multiplicative largeness of the set $\mathbf{POL}.$ That’s why we postpone our discussions up to the subsection 1.3. Before that, we recall some necessary results on the set $\mathbb{P}-\mathbb{P}.$

1.2 A brief introduction to $\mathbb{P}-\mathbb{P}$

In $1905,$ Maillet [M05] conjectured the following conjecture.

Conjecture 1.3.

[M05] Every even number is the difference of two primes.

Originally before Maillet, there were two stronger forms of this conjecture. In $1901$, Kronecker [K01] made the following conjecture.

Conjecture 1.4.

[K01] Every even number can be expressed in infinitely many ways as the difference of two primes.

In $1849$, Polignac [P49] conjectured the following which is the most general one.

Conjecture 1.5.

[P49] Every even number can be written in infinitely many ways as the difference of two consecutive primes. In other words, he conjectured that $\mathbf{POL}=2\mathbb{N}.$

Based on [GPY09], Zhang [Z14] made a recent breakthrough and proved that there exists an even number not more than $7\times 10^{7}$ which can be expressed in infinitely many ways as the difference of two primes. Soon after, Maynard and Tao [P14, M15] reduced the limit of such an even number to not more than $600$. The best-known result now is not more than $246$; for details see [P14]. In other words, the best-known result is $\mathbf{POL}\cap[2,246]\neq\emptyset.$

1.2.1 The Banks–Freiberg–Turnage-Butterbaugh theorem

An ordered tuple $\mathcal{H}$ of distinct non negative integers is said to be admissible if it avoids at least one residue class mod $p$ for every prime $p$. Following Tao and Ziegler [TZ23],we say that a finite admissible tuple $\mathcal{H}=\{h_{1},h_{2},\ldots,h_{k}\}$ is prime-producing if there are infinitely many $n\in\mathbb{N}$ such that $\{n+h_{1},n+h_{2},\ldots,n+h_{k}\}$ are simultaneously prime. The Dickson-Hardy- Littlewood conjecture asserts that every such tuple $\mathcal{H}$ is prime-producing. This conjecture remains one of the great unsolved problems in number theory, and the strongest unconditional result in this direction is the following theorem of Maynard [M15] and Tao.

Theorem 1.6 (Maynard-Tao).

For every integer $m>2$, there is a number $k_{m}$ for which the following holds. If $(h_{1},\ldots,h_{k})$ is admissible with $k>k_{m}$ , then the set $\{n+h_{1},\ldots,n+h_{k}\}$ contains at least $m$ primes for infinitely many $n\in\mathbb{N}.$

Soon after the announcement of the above theorem, Banks–Freiberg–Turnage-Butterbaugh [BFT15] improved the above theorem and solved an old conjecture of P. Erdős. Let us recall their theorem.

Theorem 1.7 (Banks–Freiberg–Turnage-Butterbaugh).

Fix an integer $m>2$, and let $k_{m}$ have the property stated in Theorem 1.6. If $(h_{1},\ldots,h_{k})$ is admissible with $k>k_{m}$, then there is a set $\{h^{\prime}_{1},\ldots,h^{\prime}_{m}\}\subseteq\{h_{1},\ldots,h_{k}\}$ such that the set $\{n+h^{\prime}_{1},\ldots,n+h^{\prime}_{m}\}$ consists of $m$ consecutive primes for infinitely many $n\in\mathbb{N}$.

1.3 A brief review of the algebra of the Stone-Čech compactification of discrete semigroups

In this subsection, we recall some basic preliminaries of the algebra of the ultrafilters. For details the readers can see the beautiful book on the algebra of ultrafilters [HS12] and a short review [BBDiNJ08, Chapter 2]. Let $(S,\cdot)$ be a discrete semigroup. Denote by $\beta S$, The collection of all ultrafilters is over $S$. For any $A\subseteq S,$ define $\overline{A}=\{p:A\in p\}.$ The collection $\{\overline{A}:A\subseteq S\}$ forms a basis, and generate a topology over $\beta S$ under which $\beta S$ becomes compact Hausdorff. It can be shown that $\beta S$ is the Stone-Čech compactification of $S$. For any $p,q\in\beta S$, define $p\cdot q\in\beta S$ as $A\in p\cdot q$ if and only if $\{s:s^{-1}A\in q\}\in p,$ where $s^{-1}A=\{t:s\cdot t\in A\}.$ It can be proved that with this operation, $\beta S$ becomes a compact, right topological semigroup. In [E58], Ellis proved that every compact right topological semigroup contains idempotents. In fact, it can be shown that a set $A$ contains an $IP$ set if and only if $A\in p$ for some idempotent $p\in\beta S$. From [BBDiNJ08, Chapter 2, Lemma 4.4 (ii)], we know that $A$ is $IP^{\star}$ set if and only if $A\in p$ for every idempotents in $\beta S$.

A set $L\subseteq\beta S$ is called a left ideal if $\beta S\cdot L\subseteq L.$ An equivalent formulation ([HS12, Theorem 4.48]) of syndetic sets says that $A\subseteq S$ is syndetic if and only if $\overline{A}\cap L\neq\emptyset$ for every left ideal $L$ of $\beta S$. Using Zorn’s lemma one can show that every left ideal contains minimal ideals. Let $K(\beta S,\cdot)$ be the union of all minimal left ideals.

Remark 1.8.

One can show that every left ideal contains idempotents. Hence every $IP^{\star}$ set is syndetic. As the set of odd numbers does not contain any $IP$ set in $(\mathbb{N},+)$, $2\mathbb{N}\in p$ for every idempotent $p\in(\beta\mathbb{N},+).$ That means for every $IP$ set $A$, $A\cap 2\mathbb{N}\in p$ for some idempotents, and so contains an $IP$ set. This immediately implies the following strengthening of Theorem 1.2.

Theorem 1.9 (Improved Banks theorem).

The set $\mathbf{POL}$ is an $IP^{\star}$ set, hence an additively syndetic set.

In this article we show that the set $\mathbf{POL}$ is also multiplicative syndetic. In other words there exists a finite set $E$ such that $\mathbb{N}=\bigcup_{s\in E}s^{-1}\mathbf{POL}.$ In fact our result is so general than Theorem 1.9 that this implies the set $\mathbf{POL}$ is both additive and multiplicative syndetic.

From [HS12, Theorem 4.40], we know that $A\subseteq S$ is piecewise syndetic if and only if $\overline{A}\cap K(\beta S,\cdot)\neq\emptyset.$ Using Folkman-Sander theorem [GR71], it is easy to verify that every multiplicative piecewise syndetic subset of $\mathbb{N}$ contains additive $IP_{n}$ sets for every $n\in\mathbb{N}$. In other words, for every $n\in\mathbb{N},$ every $IP_{n}^{\star}$ set belongs to every $p\in K(\beta S,\cdot),$ in other words every $IP_{n}^{\star}$ set is syndetic. The following theorem is our main result.

Theorem 1.10 (Main theorem).

There exists $N\in\mathbb{N}$ such that the set $\mathbf{POL}$ is $IP_{N}^{\star},$ hence $K(\beta\mathbb{N},\cdot)\subseteq\overline{\mathbf{POL}}.$ Hence $\mathbf{POL}$ is a multiplicative syndetic set.

2 Proof of Our results

Now we are in the position to prove our main theorem. Till now we have reduced our main problem to a simple combinatorial problem stating that the set $\mathbf{POL}$ intersects every $IP_{N}$ set for some sufficiently large number $N$. Now we use the art of pigeonhole principle to solve the reduced problem.

Proof of Theorem 1.10:.

Let us choose $m=2$, and let $k=k_{2}$ be the number coming from Theorem 1.7. Choose a sufficiently large number $N\in\mathbb{N}$ such that we can do all of our following calculations. We will apply the Pigeonhole principle iteratively. Our number $N$ depends only on the first $k$ primes, and the Pigeonhole principle applied $k$ times. So this $N$ is computable, but too high to calculate. Let us enumerate the set of primes $\mathbb{P}$ as $(p_{n})_{n}.$

To show that the set $\mathbf{POL}$ is $IP_{N}^{\star},$ we need to show that for any given $IP_{N}$ set $FS(\langle x_{n}\rangle_{n=1}^{N})$ in $\mathbb{N},$ $\mathbf{POL}\cap FS(\langle x_{n}\rangle_{n=1}^{N})\neq\emptyset$. We will show that there exists $1\leq i<j\leq N$ such that $\mathbf{POL}\cap(x_{i}+\cdots+x_{j})\neq\emptyset$. To verify this, arbitrarily choose any $IP_{N}$ set $FS(\langle x_{n}\rangle_{n=1}^{N})\subseteq\mathbb{N}$.

To proceed inductively, consider the following subset of $FS(\langle x_{n}\rangle_{n=1}^{N})\subseteq\mathbb{N}$

$$C_{1}=\{x_{1},x_{1}+x_{2},\ldots,x_{1}+x_{2}+\cdots+x_{N}\}.$$

Applying the pigeonhole principle, there exists $h_{1}\in[0,p_{1}-1]$ such that $C_{2}\cap(p_{1}\mathbb{N}+h_{1})$ has $N_{1}$ elements, which is large enough for the next steps. Now

$$C_{2}=\{x_{1}+\cdots+x_{i_{1}},x_{1}+\cdots+x_{i_{2}},\ldots,x_{1}+\cdots+x_{i_{N_{1}}}\}\subset C_{1}$$

for some sequence $\{i_{1}<i_{2}<\cdots<i_{N_{1}}\}\subset\{1,2,\ldots,N\}.$ Note that the number $N_{1}$ depends on the pigeonhole principle and the first prime $p_{1}.$ Define

• $b_{1}=x_{1}+\cdots+x_{i_{1}}\in C_{2},$ and

• $C^{\prime}_{2}=C_{2}\setminus\{b_{1}\}.$

Now apply the above argument to extract $h_{2}\in[0,p_{2}-1]$ and an another set $C_{3}=C^{\prime}_{2}\cap(p_{2}\mathbb{N}+h_{2})$, where

$$C_{3}=\{x_{1}+\cdots+x_{j_{1}},x_{1}+\cdots+x_{j_{2}},\ldots,x_{1}+\cdots+x_{j_{N_{2}}}\}\subset C_{2},$$

and $\{j_{1}<j_{2}<\cdots<j_{N_{2}}\}\subset\{i_{1},i_{2},\ldots,i_{N_{1}}\}.$ Note that the number $N_{2}$ depends on the pigeonhole principle and the second prime $p_{2}.$ Now define

• $b_{2}=x_{1}+\cdots+x_{j_{1}},$ and

• $C^{\prime}_{3}=C_{3}\setminus\{b_{2}\}.$

Now apply this argument $k$ times to extract the elements $\{b_{1},b_{2},\ldots,b_{k}\}\subset FS(\langle x_{n}\rangle_{n=1}^{N})$. And also $b_{i}\in C_{j}$ if $k\geq i\geq j.$ But note that for every $n\in\mathbb{N}$, $n<p_{n}.$ Hence

$$|\{b_{1},b_{2},\ldots,b_{k}\}(\!\!\!\mod p_{n})|\leq n<p_{n},$$

showing that the sequence $B=\{b_{1},b_{2},\ldots,b_{k}\}$ is admissible.

Using Theorem 1.7, we can choose $n\in\mathbb{N}$, and $x,y(>x)\in B$ such that $n+x,n+y$ are consecutive primes. Hence $y-x\in\mathbf{POL}.$ But from the construction of the set $B$ we have $y-x\in FS(\langle x_{n}\rangle_{n}).$ Which implies $\mathbf{POL}\cap FS(\langle x_{n}\rangle_{n=1}^{N})\neq\emptyset.$ Hence $\mathbf{POL}$ is an $IP_{N}^{\star}$ set. ∎

Remark 2.1.

From [G23, Theorem 1.14], it follows that there exists $k\in\mathbb{N}$ such that $k\cdot\mathbb{N}\subseteq\mathbf{POL}\cdot\mathbf{POL}.$

3 Concluding remarks

In this section, we address a possible question that appears immediately after our main theorem 1.10. First, we need the following notion of largeness arising from the difference of sets.

Definition 3.1 ($\Delta_{r}$-set and $\Delta_{r}^{*}$-set).

Let $r$ be a given positive integer.

• For $S\subset\mathbb{N}$ with $|S|\geq r$, its difference set

  $$\Delta(S)=(S-S)\cap\mathbb{N}=\{a-b:a,b\in S,a>b\}$$

  is known as a $\Delta_{r}$-set.

• A set $S\subset\mathbb{N}$ is called a $\Delta_{r}^{*}$-set if the intersection of $S$ with any $\Delta_{r}$-set is not empty.

An even number $n$ is called a Maillet number (Kronecker number), if it can be written (in infinitely many ways) as the difference of two primes. Let $\mathcal{K}$ be the set of all Kronecker numbers. In [HW17], Huang and Wu proved $\mathcal{K}$ is a $\Delta_{r}^{*}$-set.

Theorem 3.2.

$\mathcal{K}$ is a $\Delta_{r}^{*}$-set for any $r\geq 721$.

It is easy to verify that every $\varDelta_{r}^{\star}$ set is $IP_{r}^{\star}$ (see [G23, Page: 2]). So it is natural to ask for a strengthening of Theorem 1.10. We believe that the answer to the following question should be affirmative.

Question 3.3.

Does the set $\mathbf{POL}$ is a $\Delta_{r}^{*}$ for some $r\in\mathbb{N}?$