Abstract:For $h=3$ and $h=4$ we prove the existence of infinite $B_h$ sequences $\B$ with counting function
$$\mathcal{B}(x)= x^{\sqrt{(h-1)^2+1}-(h-1) + o(1)}.$$ This result extends a construction of I. Ruzsa for $B_2$ sequences.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to construct an infinite \(B_h\) sequence with specific growth properties. Specifically, the author hopes to prove that for \(h = 3\) and \(h = 4\), there exists an infinite \(B_h\) sequence \(B\) whose counting function is:
\[ B(x) = x^{\sqrt{(h - 1)^2+ 1}-(h - 1)+o(1)} \]
This result extends Ruzsa's construction method for \(B_2\) sequences. The following is a detailed explanation of this problem:
### Research Background
A \(B_h\) sequence refers to a sequence \(B\) of positive integers in which all sums composed of \(h\) elements (which can be the same) are distinct. That is, for any \(b_1, b_2,\ldots, b_h\in B\) and \(b'_1, b'_2,\ldots, b'_h\in B\), if \(b_1\leq b_2\leq\cdots\leq b_h\) and \(b'_1\leq b'_2\leq\cdots\leq b'_h\), then:
\[ b_1 + b_2+\cdots+ b_h\neq b'_1 + b'_2+\cdots+ b'_h \]
Studying the size of a finite \(B_h\) set or the growth function of an infinite \(B_h\) sequence is a classic topic in combinatorial number theory. Define:
\[ F_h(n)=\max\{|B|: B\text{ is }B_h, B\subset[1, n]\} \]
Existing research has shown that for a constant \(C_h\), there is an inequality:
\[ F_h(n)\leq(C_h + o(1))n^{1/h} \]
Therefore, when \(B\) is an infinite \(B_h\) sequence, we have:
\[ B(x)\ll x^{1/h} \]
### Core of the Problem
Erdős conjectured that for any \(\epsilon> 0\), there exists an infinite \(B_h\) sequence \(B\) whose counting function satisfies:
\[ B(x)\gg x^{1/h-\epsilon} \]
But it is currently impossible to prove whether \(\epsilon\) can be removed, especially for the case where \(h\geq3\).
The known largest growth index is \(1/(2h - 1)\), which is constructed by a greedy algorithm. For \(h = 2\), Atjai, Komlós and Szemerédi proposed an improved lower bound, and Ruzsa constructed an infinite Sidon sequence (i.e., a \(B_2\) sequence) by a clever method, whose counting function is:
\[ B(x)=x^{\sqrt{2}-1 + o(1)} \]
### Objectives of the Paper
This paper aims to apply Ruzsa's ideas to the cases of \(h = 3\) and \(h = 4\) in order to improve the growth lower bound when \(h\geq3\). The specific objective is to prove that there exists an infinite \(B_h\) sequence \(B\) whose counting function is:
\[ B(x)=x^{\sqrt{(h - 1)^2+ 1}-(h - 1)+o(1)} \]
This will be an important progress because the previous best lower bound was \(x^{1/(2h - 1)}\), and the new lower bound significantly improves the growth rate.