Polynomial bound for x₁+x₂+2x₃=x₄+3x₅ sets

Improvement of the bound in Theorem 1.12 · arXiv:2004.10180

arXiv Informal medium confidence— first stated 2021-09-25

Status open high confidence

No follow-up work has been found that improves the bound from $o(\sqrt{n})$ to $n^{1/2-\varepsilon}$ for subsets of $[n]$ without a nontrivial solution to $x_1+x_2+2x_3=x_4+3x_5$. Prendiville (arXiv:2005.03484, 2020) gave an alternative proof of the $o(\sqrt{n})$ result of Conlon--Fox--Sudakov--Zhao with effective (quantitative) bounds and extensions to equations in more variables, but this work predates the 2021 publication and does not establish a polynomial gain. The open question of whether $n^{1/2-\varepsilon}$ holds for some $\varepsilon>0$ remains unresolved.

Reviewer notes. Prendiville (arXiv:2005.03484, 2020, published Math. Proc. Cambridge Phil. Soc. 2022) provides an alternative proof of the o(sqrt(n)) result and effective bounds, but predates the paper's formal 2021 publication and does not address the polynomial improvement question. No papers after 2021 found that resolve or make progress on the n^{1/2-epsilon} question.

Auto-reviewed 2026-05-15 with claude-sonnet-4-6 (web search enabled).

Informal. Can the bound $o(\sqrt{n})$ for subsets of $[n]$ without a nontrivial solution to $x_1+x_2+2x_3=x_4+3x_5$ be improved to $n^{1/2-\varepsilon}$ for some $\varepsilon>0$?

Context

Theorem 1.12 shows that every subset of $[n]$ without a nontrivial solution to $x_1+x_2+2x_3=x_4+3x_5$ has size $o(\sqrt{n})$, improving on the trivial Sidon-set upper bound of $(1+o(1))\sqrt{n}$. The authors explicitly identify as an open problem whether the bound can be further strengthened to $n^{1/2-\varepsilon}$.

Source paper

The regularity method for graphs with few 4-cycles
David Conlon, Jacob Fox, Benny Sudakov, Yufei Zhao · 2021-09-25
https://arxiv.org/abs/2004.10180 PDF source