1. Introduction
Let
$\mathbb{N}$ be the set of all non-negative integers. For any integer r and m, let
$r+m\mathbb{N}=\{r+mk: k\in\mathbb{N}\}$. Let A be the set of all non-negative integers which contain an even number of digits 1 in their binary representations and
$B=\mathbb{N}\backslash A$. The set A is called Thue–Morse sequence. For any positive integer l, let
$A_{l}=A\cap [0, 2^{l}-1]$ and
$B_{l}=B\cap [0, 2^{l}-1]$. For
$S\subseteq \mathbb{N}$ and
$n\in \mathbb{N}$, let the representation function
$R_{S}(n)$ denote the number of solutions of the equation
$s+s'=n$ with
$s, s'\in S$ and
$s \lt s'$. Sárközy asked whether there exist two subsets
$C, D\subseteq \mathbb{N}$ with
$|(C\cup D)\backslash (C\cap D)|=\infty$ such that
$R_{C}(n)=R_{D}(n)$ for all sufficiently large integers n. By using the Thue–Morse sequence, Dombi [Reference Dombi6] answered Sárközy’s problem affirmatively. Later, Lev [Reference Lev10], Sándor [Reference Sándor12] and Tang [Reference Tang17] proved this result by different methods. Partitions of non-negative integers and their corresponding representation functions have been extensively studied by many authors. The related results can be found in [Reference Chen and Tang4, Reference Chen and Wang5, Reference Jiao, Sándor, Yang and Zhou7–Reference Kiss and Sándor9, Reference Sándor12–Reference Tang17, Reference Yang and Chen19].
In 2012, Yu and Tang [Reference Yu and Tang20] began to focus on partitions of non-negative integers with the intersection not empty. They studied the intersection of two sets is an arithmetic progression and posed the following conjecture:
Conjecture 1.1.
Let
$m\in\mathbb{N}$ and
$R\subset \{0, 1, \ldots, m-1\}$. If
$C\cup D=\mathbb{N}$ and
$C\cap D=\{r+km: k\in\mathbb{N}, r\in R\}$, then
$R_{C}(n)=R_{D}(n)$ cannot hold for all sufficiently large n.
In 2016, Tang [Reference Tang18] obtained the following theorem.
Theorem A [Reference Tang18, Theorem 1]
Let m be an integer with
$m\geq 2$. If
$C\cup D=\mathbb{N}$ and
$C\cap D=m\mathbb{N}$, then
$R_{C}(n)=R_{D}(n)$ cannot hold for all large enough integers n.
In 20l6, Chen and Lev [Reference Chen and Lev3] disproved Conjecture 1.1 by the following result.
Theorem B [Reference Chen and Lev3, Theorem 1]
Let l be a positive integer. There exist two sets C and D with
$C\cup D=\mathbb{N}$ and
$C\cap D=(2^{2l}-1)+(2^{2l+1}-1)\mathbb{N}$ such that
$R_{C}(n)=R_{D}(n)$ for every positive integer n.
In [Reference Chen and Lev3], Chen and Lev also proposed the following problem.
Problem 1.2.
Given
$R_{C}(n)=R_{D}(n)$ for every positive integer n,
$C\cup D=\mathbb{N}$ and
$C\cap D=r+m\mathbb{N}$ with
$r\geq 0$ and
$m\geq 2$, must there exist an integer
$l\geq 1$ such that
$r=2^{2l}-1, m=2^{2l+1}-1$?
Afterwards, Li and Tang [Reference Li and Tang11], Chen, Tang and Yang [Reference Chen, Tang and Yang2] solved Problem 1.2 under the condition
$0\leq r \lt m$. In 2021, Chen and Chen [Reference Chen and Chen1] solved Problem 1.2 affirmatively.
Theorem C [Reference Chen and Chen1, Theorem 1.1]
Let
$m\geq 2$ and
$r\geq 0$ be two integers and let C and D be two sets with
$C\cup D=\mathbb{N}$ and
$C\cap D=r+m\mathbb{N}$ such that
$R_{C}(n)=R_{D}(n)$ for every positive integer n. Then there exists a positive integer l such that
$r=2^{2l}-1$ and
$m=2^{2l+1}-1$.
Let
$r_{1}, r_{2}, m$ be integers with
$0 \lt r_{1} \lt r_{2} \lt m$. In this paper, we focus on partitions of non-negative integers into two sets
$C, D$ with
$C\,\cup\,D=\mathbb{N}$ and
$C\,\cap\,D=(r_{1}\,+\,m\mathbb{N})\cup (r_{2}\,+\,m\mathbb{N})$ such that
$R_{C}(n)=R_{D}(n)$ for all
$n\in\mathbb{N}$ and obtain the following result.
Theorem 1.3. Let
$r_{1}, r_{2}, m$ be integers with
$0 \lt r_{1} \lt r_{2} \lt m$ and
$2\mid r_{1}$. Then there exist two sets C and D with
$C\cup D=\mathbb{N}$ and
$C\cap D=(r_{1}+m\mathbb{N})\cup (r_{2}+m\mathbb{N})$ such that
$R_{C}(n)=R_{D}(n)$ for all
$n\in\mathbb{N}$ if and only if there exists a positive integer l such that
$r_{1}=2^{2l+1}-2, r_{2}=2^{2l+1}-1, m=2^{2l+2}-2$.
Motivated by Theorems B and C, we propose the following conjecture for further research.
Conjecture 1.4.
Let
$r_{1}, r_{2}, m$ be integers with
$0 \lt r_{1} \lt r_{2} \lt m$ and
$2\nmid r_{1}$. Then there exist two sets C and D with
$C\cup D=\mathbb{N}$ and
$C\cap D=(r_{1}+m\mathbb{N})\cup (r_{2}+m\mathbb{N})$ such that
$R_{C}(n)=R_{D}(n)$ for all
$n\in \mathbb{N}$ if and only if there exists a positive integer l such that
$r_{1}=2^{2l}-1, r_{2}=2^{2l+1}+2^{2l}-2, m=2^{2l+2}-2$.
Throughout this paper, let
$f(x)=a_{0}+a_{1}x+\cdots+a_{n}x^{n}\in\mathbb{Z}[x]$ and for
$m\leq n$, define
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU1.png?pub-status=live)
For
$C, D\subseteq \mathbb{N}$ and
$n\in \mathbb{N}$, let
$R_{C, D}(n)$ be the number of solutions of
$n=c+d$ with
$c\in C$ and
$d\in D$. Let
$C+D=\{c+d: c\in C, d\in D\}$. Let C(x) be the set of integers in C which are less than or equal to x. The characteristic function of C is denoted by
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU2.png?pub-status=live)
2. Some lemmas
Lemma 2.1. [Reference Chen and Lev3, Lemma 1]
Suppose that
$C_{0}, D_{0}\subseteq \mathbb{N}$ satisfy
$R_{C_{0}}(n)=R_{D_{0}}(n)$ for all
$n\in \mathbb{N}$, and that m is a non-negative integer with
$m \notin (C_{0}-D_{0})\cup(D_{0}-C_{0})$. Then, letting
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU3.png?pub-status=live)
we have
$R_{C_{1}}(n)=R_{D_{1}}(n)$ for all
$n\in \mathbb{N}$ and furthermore
(i)
$C_{1}\cup D_{1}=(C_{0}\cup D_{0})\cup (m+C_{0}\cup D_{0})$;
(ii)
$C_{1}\cap D_{1}\supseteq(C_{0}\cap D_{0})\cup (m+C_{0}\cap D_{0})$, the union being disjoint.
Moreover, if
$m\notin (C_{0}-C_{0})\cup (D_{0}-D_{0})$, then also in (i) the union is disjoint, and in (ii) the inclusion is in fact an equality. In particular, if
$C_{0}\cup D_{0}=[0,m-1]$, then
$C_{1}\cup D_{1}=[0,2m-1]$, and if C 0 and D 0 indeed partition the interval
$[0, m-1]$, then C 1 and D 1 partition the interval
$[0, 2m-1]$.
Lemma 2.2. [Reference Kiss and Sándor8, Claim 1]
Let
$0 \lt r_{1} \lt \cdots \lt r_{s}\leq m$ be integers. Then there exists at most one pair of sets (C, D) such that
$C\cup D=[0, m], 0\in C, C\cap D=\{r_{1}, \ldots, r_{s}\}$ and
$R_{C}( n)=R_{D}(n)$ for every
$n \leq m$.
Lemma 2.3. [Reference Kiss and Sándor8, Claim 3]
If for some positive integer M, the integers
$M-1, M-2, M-4, M-8, \ldots, M-2^{\lceil \log_{2}M\rceil-1}$ are all contained in the set A, then
$\lceil \log_{2}M\rceil$ is odd and
$M=2^{\lceil \log_{2}M\rceil}-1$.
Lemma 2.4. [Reference Kiss and Sándor8, Claim 4]
If for some positive integer M, the integers
$M-1, M-2, M-4, M-8, \ldots, M-2^{\lceil \log_{2}M\rceil-1}$ are all contained in the set B, then
$\lceil \log_{2}M\rceil$ is even and
$M=2^{\lceil \log_{2}M\rceil}-1$.
Lemma 2.5. [Reference Kiss and Sándor8, Theorem 3]
Let C and D be sets of non-negative integers such that
$C\cup D=[0, m], C\cap D=\emptyset$ and
$0\in C$. Then
$R_{C}(n)=R_{D}(n)$ for every positive integer n if and only if there exists a positive integer l such that
$C=A_{l}$ and
$D=B_{l}$.
3. Proofs
Proof of Theorem 1.3. (Sufficiency). For any given positive integer l, let
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn1.png?pub-status=live)
For given sets
$C_{0}=\{0\}, D_{0}=\{1\}$, define
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn2.png?pub-status=live)
and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn3.png?pub-status=live)
It is clear that C 0 and D 0 partition the interval
$[0, m_{0}-1]$ and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU4.png?pub-status=live)
and
$R_{C_{0}}(n)=R_{D_{0}}(n)$ for all
$n\in \mathbb{N}$ (both representation functions are identically equal to 0). Applying Lemma 2.1 inductively
$2l-1$ times, we can deduce that for any
$i\in [0, 2l-1]$,
$R_{C_{i}}(n)=R_{D_{i}}(n)$ for all
$n\in\mathbb{N}$, the sets Ci and Di partition the interval
$[0, m_{i}-1]$ and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU5.png?pub-status=live)
In particular,
$R_{C_{2l-1}}(n)=R_{D_{2l-1}}(n)$ for all
$n\in \mathbb{N}$, the sets
$C_{2l-1}$ and
$D_{2l-1}$ partition the interval
$[0, m_{2l-1}-1]=[0, 2^{2l}-1]$ and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU6.png?pub-status=live)
By Lemma 2.1, we have
$R_{C_{2l}}(n)=R_{D_{2l}}(n)$ for all
$n\in \mathbb{N}$, the sets
$C_{2l}$ and
$D_{2l}$ partition the interval
$[0, 2m_{2l-1}-1]=[0, 2^{2l+1}-1]=[0, m_{2l}+1]$. In addition, it is easily seen that
$\{0, m_{2l}\}\subseteq C_{2l}$ and
$\{1, m_{2l}+1\}\subseteq D_{2l}$. Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU7.png?pub-status=live)
By Lemma 2.1, we have
$R_{C_{2l+1}}(n)=R_{D_{2l+1}}(n)$ for all
$n\in \mathbb{N}$ and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU8.png?pub-status=live)
Applying again Lemma 2.1, we can conclude that
$R_{C_{i}}(n)=R_{D_{i}}(n)$ for all
$n\in \mathbb{N}$,
$C_{i}\cup D_{i}=[0, m_{i}-1]$ and
$C_{i}\cap D_{i}=\{m_{2l}, m_{2l}+1\}+\{0, m_{2l+1}, \ldots, (2^{i-2l}-1)m_{2l+1}\}$ for each
$i\geq 2l+1$.
Therefore, by the definitions of C and D in (3.1)–(3.3), we have
$R_{C}(n)=R_{D}(n)$ for all
$n\in \mathbb{N}$,
$C\cup D=\mathbb{N}$ and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU9.png?pub-status=live)
(Necessity). To prove the necessity of Theorem 1.3, we need the following three claims.
Claim 1.
Given
$0 \lt r_{1} \lt r_{2}\leq m$, there exists at most one pair of sets (C, D) such that
$C\cup D=\mathbb{N}$,
$C\cap D=(r_{1}+m\mathbb{N})\cup (r_{2}+m\mathbb{N})$ and
$R_{C}(n)=R_{D}(n)$ for all
$n\in\mathbb{N}$.
Proof of Claim 1
Assume that there exist at least two pairs of sets (C, D) and
$(C', D')$ which satisfy the conditions
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU10.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU11.png?pub-status=live)
We may assume that
$0\in C\cap C'$. Let k be the smallest positive integer such that
$\chi_{C}(k)\neq\chi_{C'}(k)$. Write
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU12.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU13.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU14.png?pub-status=live)
Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn4.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn5.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn6.png?pub-status=live)
For any integer
$n\in [0, k]$, by the hypothesis, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn7.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn8.png?pub-status=live)
Thus there exist two pairs of sets
$(C_{1}, D_{1})$ and
$(C_{2}, D_{2})$ satisfying (3.4)–(3.8). By Lemma 2.2, this is impossible. This completes the proof of Claim 1.
Claim 2.
Let
$r_{1}, r_{2}, m$ be integers with
$0 \lt r_{1} \lt r_{2} \lt r_{1}+r_{2}\leq m$ and
$2\mid r_{1}$. Let C and D be sets of non-negative integers such that
$C\cup D=[0, m]$,
$C\cap D=\{r_{1}, r_{2}\}$ and
$0\in C$. If
$R_{C}(n)=R_{D}(n)$ for any integer
$n\in [0, m]$, then there exists a positive integer l such that
$r_{1}=2^{2l+1}-2, r_{2}=2^{2l+1}-1$.
Proof of Claim 2
Let
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn9.png?pub-status=live)
Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn10.png?pub-status=live)
Since
$R_{C}(n)=R_{D}(n)$ for any integer
$n\in [0, m]$, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn11.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU15.png?pub-status=live)
Noting that
$C\cup D=[0, m]$,
$C\cap D=\{r_{1}, r_{2}\}$, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU16.png?pub-status=live)
Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU17.png?pub-status=live)
Thus
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn12.png?pub-status=live)
An easy calculation shows that
$r_{1}\geq 6$,
$\{0, 3, 5, 6\}\subset C$ and
$\{1, 2, 4, 7\}\subset D$.
In order to prove
$r_{2}=r_{1}+1$, we suppose that
$r_{2}\geq r_{1}+2$ and we will show that this leads to a contradiction.
The coefficient of
$x^{r_{1}-1}$ in (3.12) is
$0=2\sum\limits_{i=0}^{r_{1}-1}\chi_{C}(i)-r_{1}$. Since
$r_{1}\in C$, we have
$\chi_{C}(r_{1})=1$. Then
$2\sum\limits_{i=0}^{r_{1}}\chi_{C}(i)=r_{1}+2$. The coefficient of
$x^{r_{1}}$ in (3.12) is
$2\chi_{C}\big(\frac{r_{1}}{2}\big)=2\sum\limits_{i=0}^{r_{1}}\chi_{C}(i)-r_{1}=2$. Then
$\chi_{C}\big(\frac{r_{1}}{2}\big)=1$. The coefficient of
$x^{r_{1}+1}$ in (3.12) is
$0=2\sum\limits_{i=0}^{r_{1}+1}\chi_{C}(i)-r_{1}-4=2\chi_{C}(r_{1}+1)-2$. Then
$\chi_{C}(r_{1}+1)=1$. The coefficient of
$x^{r_{1}+2}$ in (3.12) is
$2\chi_{C}\big(\frac{r_{1}+2}{2}\big)=2\sum\limits_{i=0}^{r_{1}+2}\chi_{C}(i)-r_{1}-4$. Then
$\chi_{C}\big(\frac{r_{1}+2}{2}\big)=\chi_{C}(r_{1}+2)$. If
$r_{2}=r_{1}+2$, then
$\chi_{C}(r_{1}+2)=1$. Comparing the coefficients of
$x^{r_{1}+s}$ with
$s\in \{3, 4, 5\}$ on the both sides of (3.12), we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU18.png?pub-status=live)
Then
$\chi_{C}(r_{1}+3)=0$,
$\chi_{C}(r_{1}+4)=1$ and
$\chi_{C}(r_{1}+5)=-1$, a contradiction. Thus
$r_{2}\geq r_{1}+3$. The coefficient of
$x^{r_{1}+3}$ in (3.12) is
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU19.png?pub-status=live)
Then
$\chi_{C}(r_{1}+2)=\chi_{C}(r_{1}+3)=0$. Thus
$\chi_{C}\big(\frac{r_{1}+2}{2}\big)=0$ and
$r_{2}\geq r_{1}+4$. The coefficient of
$x^{r_{1}+4}$ in (3.12) is
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU20.png?pub-status=live)
Then
$\chi_{C}(r_{1}+4)=1$,
$\chi_{C}\big(\frac{r_{1}+4}{2}\big)=0$ and
$2\sum\limits_{i=0}^{r_{1}+4}\chi_{C}(i)=r_{1}+6$. By Lemma 2.2, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn13.png?pub-status=live)
Since
$\chi_{C}\big(\frac{r_{1}+2}{2}\big)=\chi_{C}\big(\frac{r_{1}+4}{2}\big)$ and
$\frac{r_{1}+4}{2}\leq r_{1}-1$, by (3.13) and the definition of A, we have
$r_{1}\equiv 0\pmod 4$. It follows that
$r_{1}\geq 8$ and
$\chi_{C}\big(\frac{r_{1}+6}{2}\big)=1$.
Let k be a positive even integer such that
$r_{1}\leq k \lt k+1 \lt \min\{r_{2}, 2r_{1}\}\leq m$. Comparing the coefficients of x k and
$x^{k+1}$ on the both sides of (3.12) respectively, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU21.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU22.png?pub-status=live)
Subtracting the above two equalities and dividing by 2 we can get
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn14.png?pub-status=live)
Since
$k+1-r_{1} \lt r_{1}, k-r_{1}$ is even, by (3.13), we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU23.png?pub-status=live)
If
$\chi_{C}(k-r_{1})=0$, then
$\chi_{C}(k+1-r_{1})=1$. By (3.14), we get
$\chi_{C}(\frac{k}{2})=0$. If
$\chi_{C}(k-r_{1})=1$, then
$\chi_{C}(k+1-r_{1})=0$. By (3.14), we get
$\chi_{C}(\frac{k}{2})=1$. Thus
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn15.png?pub-status=live)
If
$\min\{r_{2}, 2r_{1}\} \gt 2r_{1}-1$, then choose
$k=2r_{1}-2^{i+1}$ with
$i\geq 0$ in (3.15), we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU24.png?pub-status=live)
Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU25.png?pub-status=live)
By Lemmas 2.3 and 2.4, we have
$r_{1}=2^{\lceil\log_{2}r_{1}\rceil}-1$, which contradicts
$2\mid r_{1}$.
If
$r_{2}=2r_{1}-1$, then compare the coefficients of
$x^{r_{2}}$ and
$x^{r_{2}-1}$ on the both sides of (3.12) respectively, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU26.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU27.png?pub-status=live)
Then
$2\chi_{C}(r_{1}-1)=\chi_{C}(r_{1}-2)$. It follows that
$\chi_{C}(r_{1}-2)=\chi_{C}(r_{1}-1)=0$, which contradicts
$\chi_{C}(r_{1}-2)+\chi_{C}(r_{1}-1)=1$. Thus
$r_{2}\leq 2r_{1}-2$.
Let k be a non-negative integer such that
$2r_{1}\leq 2r_{1}+k \lt 2r_{1}+k+1 \lt r_{1}+r_{2}\leq m$. If k is even, then compare the coefficients of
$x^{2r_{1}+s}$ with
$s\in \{k, k+1\}$ on the both sides of (3.12), we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU28.png?pub-status=live)
Subtracting the above two equalities and dividing by 2 we can get
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn16.png?pub-status=live)
If k is odd, then compare the coefficients of
$x^{2r_{1}+s}$ with
$s\in \{k, k+1\}$ on the both sides of (3.12), we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU29.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU30.png?pub-status=live)
Subtracting the above two equalities and dividing by 2 we can get
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn17.png?pub-status=live)
If r 2 is even, then choose k = 0 and k = 2 in (3.16) respectively, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU31.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU32.png?pub-status=live)
Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU33.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU34.png?pub-status=live)
By (3.13), we have
$\chi_{C}(2r_{1}-r_{2})+\chi_{C}(2r_{1}+1-r_{2})=1$ and
$\chi_{C}(2r_{1}+2-r_{2})+\chi_{C}(2r_{1}+3-r_{2})=1$. Then
$\chi_{C}(2r_{1}-r_{2})=1$ and
$\chi_{C}(2r_{1}+2-r_{2})=1$. It follows that
$r_{2}\equiv 2\pmod 4$. The coefficient of
$x^{r_{2}}$ in (3.12) is
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU35.png?pub-status=live)
and the coefficient of
$x^{r_{2}+1}$ in (3.12) is
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU36.png?pub-status=live)
Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU37.png?pub-status=live)
By
$\chi_{C}(r_{2}-r_{1})+\chi_{C}(r_{2}+1-r_{1})=1$, we have
$\chi_{C}(r_{2}-r_{1})=0$,
$\chi_{C}(r_{2}+1-r_{1})=1$. By
$r_{1}\equiv 0\pmod 4$ and
$r_{2}\equiv 2\pmod 4$, we have
$\chi_{C}(r_{2}-1-r_{1})=0$,
$\chi_{C}(r_{2}-2-r_{1})=1$. The coefficient of
$x^{r_{2}-1}$ in (3.12) is
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU38.png?pub-status=live)
Then
$2\sum\limits_{i=0}^{r_{2}-1}\chi_{C}(i)=r_{2}+2$. It follows that
$2\sum\limits_{i=0}^{r_{2}}\chi_{C}(i)=r_{2}+4$ and
$\chi_{C}\big(\frac{r_{2}}{2}\big)=1$. The coefficient of
$x^{r_{2}-2}$ in (3.12) is
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU39.png?pub-status=live)
Then
$\chi_{C}\big(\frac{r_{2}-2}{2}\big)=2-\chi_{C}(r_{2}-1)$. Thus
$\chi_{C}\big(\frac{r_{2}-2}{2}\big)=\chi_{C}(r_{2}-1)=1$. By (3.13) and
$\chi_{C}\big(\frac{r_{2}-2}{2}\big)=\chi_{C}\big(\frac{r_{2}}{2}\big)=1$, we have
$r_{2}\equiv 0\pmod 4$, a contradiction.
If r 2 is odd, then
$r_{1}+5\leq r_{2}\leq 2r_{1}-3$. The coefficient of
$x^{r_{1}+5}$ in (3.12) is
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU40.png?pub-status=live)
Then
$\chi_{C}(r_{1}+5)=0$ and so
$r_{2}\geq r_{1}+7$. The coefficient of
$x^{r_{1}+6}$ in (3.12) is
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU41.png?pub-status=live)
Then
$\chi_{C}(r_{1}+6)=\chi_{C}\big(\frac{r_{1}+6}{2}\big)=1$. By choosing k = 3 and k = 5 in (3.17) respectively, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU42.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU43.png?pub-status=live)
Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU44.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU45.png?pub-status=live)
By (3.13), we have
$\chi_{C}(2r_{1}+4-r_{2})+\chi_{C}(2r_{1}+3-r_{2})=1$ and
$\chi_{C}(2r_{1}+6-r_{2})+\chi_{C}(2r_{1}+5-r_{2})=1$. Then
$\chi_{C}(2r_{1}+3-r_{2})=\chi_{C}(2r_{1}+5-r_{2})=1$. Applying again (3.13), we have
$r_{2}\equiv 1\pmod 4$. The coefficient of
$x^{2r_{1}-2}$ in (3.12) is
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU46.png?pub-status=live)
and the coefficient of
$x^{2r_{1}-1}$ in (3.12) is
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU47.png?pub-status=live)
Subtracting the above two equalities and dividing by 2 we can obtain
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU48.png?pub-status=live)
Noting that
$\chi_{C}(r_{1}-2)+\chi_{C}(r_{1}-1)=1$ and
$\chi_{C}(2r_{1}-2-r_{2})=\chi_{C}(2r_{1}-1-r_{2})$, we have
$3\chi_{C}(r_{1}-1)=2-\chi_{C}(2r_{1}-1)$. However, it is impossible. Therefore
$r_{2}=r_{1}+1$.
The remainder of the proof is similar to the proof of [Reference Sun and Pan13, Theorem 1.1]. For the sake of completeness we give the details.
Let k be a positive even integer with
$r_{2} \lt k \lt k+1 \lt 2r_{1} \lt r_{1}+r_{2}\leq m$. Comparing the coefficients of
$x^{k-1}, x^{k}$ and
$x^{k+1}$ on the both sides of (3.12) respectively, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU49.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU50.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU51.png?pub-status=live)
Calculating the above three equalities, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn18.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn19.png?pub-status=live)
By choosing
$k=2r_{1}-2$ in (3.19), we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU52.png?pub-status=live)
Then
$\chi_{C}(r_{1}-1)=\chi_{C}(r_{1}-3)$. Thus
$r_{1}\equiv 2\pmod 4$ and
$r_{2}\equiv 3\pmod 4$.
If
$k-1-r_{2}\equiv 0\pmod 4$, then
$k-r_{1}\equiv 2\pmod 4$ and
$k\equiv 0\pmod 4$. Thus
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU53.png?pub-status=live)
Hence
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU54.png?pub-status=live)
If
$\chi_{C}(k-1-r_{2})=0$, then
$\chi_{C}(k-r_{1})=1$. By (3.18), we have
$\chi_{C}(\frac{k}{2})=1$. If
$\chi_{C}(k-1-r_{2})=1$, then
$\chi_{C}(k-r_{1})=0$. By (3.18), we have
$\chi_{C}(\frac{k}{2})=0$. Thus
$\chi_{C}\big(\frac{k}{2}\big)=\chi_{C}(k-r_{1})$ and
$\chi_{C}\big(\frac{k}{2}\big)+\chi_{C}(k-1-r_{2})=1$. Noting that
$\chi_{C}(k-1-r_{2})+\chi_{C}(k-r_{2})=1$, we have
$\chi_{C}\big(\frac{k}{2}\big)=\chi_{C}(k-r_{2})$.
If
$k-1-r_{2}\equiv 2\pmod 4$, then
$k-r_{1}\equiv 0\pmod 4$ and
$k\equiv 2\pmod 4$. By (3.18), we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU55.png?pub-status=live)
Then
$\chi_{C}\big(\frac{k-2}{2}\big)=\chi_{C}(k-2-r_{1})$. Noting that
$\chi_{C}\big(\frac{k-2}{2}\big)+\chi_{C}\big(\frac{k}{2}\big)=1$ and
$\chi_{C}(k-1-r_{2})+\chi_{C}(k-r_{2})=1$, we have
$\chi_{C}\big(\frac{k}{2}\big)=\chi_{C}(k-r_{2})$.
As a result, we can obtain
$\chi_{C}\big(\frac{k}{2}\big)=\chi_{C}(k-r_{2})$. Put
$k=2r_{2}-2^{i+1}$ with
$i\geq 0$. Then
$\chi_{C}(r_{2}-2^{i})=\chi_{C}(r_{2}-2^{i+1})$. Thus
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU56.png?pub-status=live)
By Lemma 2.3, we have
$r_{1}=2^{2l+1}-2$ and
$r_{2}=2^{2l+1}-1$ for some positive integer l.
This completes the proof of Claim 2.
Claim 3.
Let l be a positive integer and let
$E, F$ be two sets of non-negative integers with
$E\cup F=[0, 3 \cdot 2^{2l+1}-4], 0\in E$ and
$E\cap F=\{2^{2l+1}-2, 2^{2l+1}-1\}$. Then
$R_{E}(n)=R_{F}(n)$ for any integer
$n\in [0, 3\cdot 2^{2l+1}-4]$ if and only if
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU57.png?pub-status=live)
Proof of Claim 3.
We first prove the sufficiency of Claim 3. It is easy to verify that
$E\cup F=[0, 3 \cdot 2^{2l+1}-4]$,
$0\in E$ and
$E\cap F=\{2^{2l+1}-2, 2^{2l+1}-1\}$.
If
$n\in [0, 2^{2l+2}-3]$, then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU58.png?pub-status=live)
and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU59.png?pub-status=live)
By Lemma 2.5, for all
$k\in\mathbb{N}$, we have
$R_{A_{2l+1}}(k)=R_{B_{2l+1}}(k)$. Then
$R_{E}(n)=R_{F}(n)$.
If
$n\in [2^{2l+2}-2, 3\cdot 2^{2l+1}-5]$, then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU60.png?pub-status=live)
and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU61.png?pub-status=live)
By Lemma 2.5,
$R_{A_{2l+1}}(k)=R_{B_{2l+1}}(k)$ holds for all
$k\in\mathbb{N}$ and then
$R_{E}(n)=R_{F}(n)$.
By
$3\cdot 2^{2l+1}-4=(2^{2l+1}-2)+(2^{2l+2}-2)$ in D, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU62.png?pub-status=live)
and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU63.png?pub-status=live)
Thus
$R_{C}(3\cdot 2^{2l+1}-4)=R_{D}(3\cdot 2^{2l+1}-4)$.
The necessity of Claim 3 follows from Lemma 2.2 and the sufficiency of Claim 3.
This completes the proof of Claim 3.
Now let
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU64.png?pub-status=live)
Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU65.png?pub-status=live)
Moreover, for any integer
$n\in [0, m-1+r_{1}]$, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU66.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU67.png?pub-status=live)
Thus for any integer
$n\in [0, m-1+r_{1}]$, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU68.png?pub-status=live)
Noting that
$r_{2}\leq m-1$, we see that
$r_{1}+r_{2}\leq m-1+r_{1}$. By Claim 2, there exists a positive integer l such that
$r_{1}=2^{2l+1}-2, r_{2}=2^{2l+1}-1$.
Let E and F be as in Claim 3. If
$m\geq 2^{2l+2}-1$ and
$0\in C$, then
$m-1+r_{1}\geq 3\cdot 2^{2l+1}-4$ and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU69.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU70.png?pub-status=live)
Moreover,
$R_{C(3 \cdot 2^{2l+1}-4)}(n)=R_{C}(n)=R_{D}(n)=R_{D(3 \cdot 2^{2l+1}-4)}(n)$ for all
$n\in [0, 3 \cdot 2^{2l+1}-4]$. By Lemma 2.2, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU71.png?pub-status=live)
By
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU72.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU73.png?pub-status=live)
we know that
$R_{C}(3\cdot 2^{2l+1}-3)=R_{D}(3\cdot 2^{2l+1}-3)$ if and only if
$\chi_{C}(3\cdot 2^{2l+1}-3)=0$, that is,
$3\cdot 2^{2l+1}-3 \in D$. Noting that
$2^{2l+1}-2\in A_{2l+1}, 2^{2l+1}-1\in B_{2l+1}$,
$3\cdot 2^{2l+1}-2=(2^{2l+1}-1)+(2^{2l+2}-1)$ in C and
$3\cdot 2^{2l+1}-2=1+(3\cdot 2^{2l+1}-3)$ in D, we obtain
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU74.png?pub-status=live)
and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU75.png?pub-status=live)
Thus by Lemma 2.5, we have
$R_{C}(3\cdot 2^{2l+1}-2) \gt R_{D}(3\cdot 2^{2l+1}-2)$, which is impossible. Therefore
$m\leq 2^{2l+2}-2$.
Now we assume that
$2^{2l+1} \leq m\leq 2^{2l+2}-3$ and
$0\in C$. Let
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU76.png?pub-status=live)
Since
$2^{2l+2}-2\leq M\leq 3\cdot 2^{2l+1}-5$, by Lemma 2.2, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn20.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn21.png?pub-status=live)
Moreover,
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn22.png?pub-status=live)
Since
$R_{C}(n)=R_{D}(n)$ for all
$n\in\mathbb{N}$ and
$0\not\in D$, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn23.png?pub-status=live)
for any integer
$n\in [0, M]$. By (3.20)–(3.23) and Lemma 2.2, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn24.png?pub-status=live)
Then
$\chi_{E}(M)=1$,
$\chi_{F}(M)=0$.
By
$2^{2l+1}-3\in A_{2l+1}$, we have
$3\,\cdot\,2^{2l+1}-5\in F$. Then
$M \lt 3\,\cdot\,2^{2l+1}-5$. If
$\chi_{E}(M+1)=1$, then
$\chi_{F}(M+1)=0$ and
$C(M+1)=E(M+1), D(M+1)=F(M+1)\cup \{M, M+1\}$. Thus
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU77.png?pub-status=live)
and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU78.png?pub-status=live)
By Claim 3, we have
$R_{E(M+1)}(M+1)=R_{F(M+1)}(M+1)$. Then
$R_{C}(M+1)\neq R_{D}(M+1)$, a contradiction. Thus
$\chi_{E}(M+1)=0$ and
$\chi_{F}(M+1)=1$.
Let t be an arbitrary positive integer such that
$M \lt M+t \lt M+t+1\leq 3\cdot 2^{2l+1}-4$. Then
$1\leq t\leq 2^{2l+1}-3$. Define the sets S and T by
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU79.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU80.png?pub-status=live)
Noting that
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU81.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU82.png?pub-status=live)
we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU83.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU84.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU85.png?pub-status=live)
Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU86.png?pub-status=live)
It follows that
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn25.png?pub-status=live)
For
$M+t\leq n\leq 3\cdot 2^{2l+1}-4$, let
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU87.png?pub-status=live)
We claim that
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn26.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn27.png?pub-status=live)
In fact, if
$E(M+t)\backslash C(M+t)=\emptyset$, then
$N_{1}(t, n)=N_{2}(t, n)=0$; if
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU88.png?pub-status=live)
for some positive integer u, then by (3.24), we have
$c_{i}\geq M+1$ and so
$0\leq n-c_{i}\leq 2^{2l+1}-3$ for
$i\in [1, u]$. In view of
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU89.png?pub-status=live)
we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU90.png?pub-status=live)
Thus (3.26) holds. Similarly, we can deduce (3.27) holds.
By
$M+t \lt 3\cdot 2^{2l+1}-4 \lt 2^{2l+3}-4\leq 2M$, we can obtain
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU91.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn28.png?pub-status=live)
Similarly, we can get
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn29.png?pub-status=live)
By choosing
$n=M+t$ and
$n=M+t+1$ in (3.28) respectively, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn30.png?pub-status=live)
and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn31.png?pub-status=live)
By choosing
$n=M+t$ and
$n=M+t+1$ in (3.29) respectively, we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn32.png?pub-status=live)
and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn33.png?pub-status=live)
Note that
$R_{C(n)}(n)=R_{D(n)}(n)$ and
$R_{E(n)}(n)=R_{F(n)}(n)$. By (3.30)–(3.33), we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU92.png?pub-status=live)
and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU93.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU94.png?pub-status=live)
and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU95.png?pub-status=live)
Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqn34.png?pub-status=live)
If M is even, then we can write
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU96.png?pub-status=live)
where
$b_{i}\in \{0, 1\}$. It follows from
$\chi_{F}(M)=0$ that
$\chi_{B_{2l+1}}\big(\sum\limits_{i=1}^{2l}b_{i}2^{i}\big)=1$. By choosing
$M+t+1=3\cdot 2^{2l+1}-4$ in (3.34), we see that t is odd and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU97.png?pub-status=live)
Then
$\chi_{E}(t+1)=0$. It follows from
$\chi_{F}(t)+\chi_{E}(t)=1$ and
$\chi_{E}(3\cdot 2^{2l+1}-4)=1$ that
$\chi_{E}(t-1)=0$ and
$\chi_{F}(t-1)=1$. Since
$\chi_{E}(t-1)+\chi_{E}(t)=1$, we have
$\chi_{E}(t)=1$ and
$\chi_{F}(t)=0$. Noting that
$\chi_{E}(t-1)=\chi_{E}(t+1)$, we have
$t\equiv 3\pmod 4$ and so
$t\geq 3$. Then
$\chi_{E}(t-2)=0$. By choosing
$M+(t-1)+1=3\cdot 2^{2l+1}-5$ in (3.34), we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU98.png?pub-status=live)
It follows from
$\chi_{E}(M+(t-1)+1)=\chi_{E}(3\cdot 2^{2l+1}-5)=0$ that
$\chi_{C}(M+(t-1)+1)=-1$, which is clearly false.
If M is odd, then we can write
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU99.png?pub-status=live)
where
$f\in\{0, 1, \ldots, 2l-1\}$ and
$b_{i}\in \{0, 1\}$. It follows from
$\chi_{E}(M+1)=0$ and
$\chi_{F}(M)=0$ that
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU100.png?pub-status=live)
Then f is odd. By choosing
$M+t+1=3\cdot 2^{2l+1}-4$ in (3.34), we see that t is even and
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU101.png?pub-status=live)
Then
$\chi_{E}(t+1)=0$ and
$\chi_{F}(t)=0$. Thus
$\chi_{E}(t)=1$. It follows from
$\chi_{E}(3\cdot 2^{2l+1}-4)=1$ that
$\chi_{E}(t-1)=0$ and
$\chi_{F}(t-1)=1$. Since
$\chi_{E}(t-1)=\chi_{E}(t+1)$, we have
$t\equiv 0\pmod 4$ and so
$t\geq 4$. Then
$\chi_{E}(t-2)=\chi_{E}(t-3)=1$ and
$\chi_{F}(t-2)=0$. By choosing
$M+(t-2)+1=3\cdot 2^{2l+1}-6$ in (3.34), we have
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20250131061213692-0765:S0013091524000907:S0013091524000907_eqnU102.png?pub-status=live)
It follows from
$\chi_{E}(M+(t-2)+1)=\chi_{E}(3\cdot 2^{2l+1}-6)=1$ that
$\chi_{C}(M+(t-2)+1)=2$, which is also impossible. Therefore
$m=2^{2l+2}-2$.
This completes the proof of Theorem 1.3.
Acknowledgements
This work was supported by the National Natural Science Foundation of China (Grant No. 12371003).