English

The structure of Sidon set systems

Combinatorics 2024-02-20 v2 Number Theory

Abstract

A family F2G\mathcal{F}\subset 2^G of subsets of an abelian group GG is a Sidon system if the sumsets A+BA+B with A,BFA,B\in \mathcal{F} are pairwise distinct. Cilleruelo, Serra and the author previously proved that the maximum size Fk(n)F_k(n) of a Sidon system consisting of kk-subsets of the first nn positive integers satisfies Cknk1Fk(n)(n1k1)+nkC_k n^{k-1}\leq F_k(n) \leq \binom{n-1}{k-1}+n-k for some constant CkC_k only depending on kk. We close the gap by proving an essentially tight structural result that in particular implies Fk(n)(1o(1))(nk1)F_k(n)\geq (1-o(1))\binom{n}{k-1}. We also use this to establish a result about the size of the largest Sidon system in the binomial random family ([n]k)p\binom{[n]}{k}_p. Extensions to hh-fold sumsets for any fixed h3h\geq 3 are also obtained.

Keywords

Cite

@article{arxiv.2211.14011,
  title  = {The structure of Sidon set systems},
  author = {Maximilian Wötzel},
  journal= {arXiv preprint arXiv:2211.14011},
  year   = {2024}
}

Comments

Significant flaw in main argument as pointed out by an anonymous referee. I am unable to fix it and thus unable to prove the main result (Theorem 5). To elaborate, I am using the following argument several times: If A is a B_2h set, then A+A will be a B_h set. This is false in a strong sense even for the "base" case of h=2: For every B_4 set A, the sumset A+A will in fact never be a Sidon set

R2 v1 2026-06-28T07:12:30.650Z