English

Splits with forbidden subgraphs

Combinatorics 2025-02-05 v3

Abstract

In this note, we fix a graph HH and ask into how many vertices can each vertex of a clique of size nn can be "split" such that the resulting graph is HH-free. Formally: A graph is an (n,k)(n,k)-graph if its vertex sets is a pairwise disjoint union of nn parts of size at most kk each such that there is an edge between any two distinct parts. Let f(n,H) = \min \{k \in \mathbb N : \mbox{there is an $(n,k)$-graph $G$ such that $H\not\subseteq G$}\} . Barbanera and Ueckerdt observed that f(n,H)=2f(n, H)=2 for any graph HH that is not bipartite. If a graph HH is bipartite and has a well-defined Tur\'an exponent, i.e., ex(n,H)=Θ(nr){\rm ex}(n, H) = \Theta(n^r) for some rr, we show that Ω(n2/r1)=f(n,H)=O(n2/r1log1/rn)\Omega (n^{2/r -1}) = f(n, H) = O (n^{2/r-1} \log ^{1/r} n). We extend this result to all bipartite graphs for which an upper and a lower Tur\'an exponents do not differ by much. In addition, we prove that f(n,K2,t)=Θ(n1/3)f(n, K_{2,t}) =\Theta(n^{1/3}) for any fixed tt.

Keywords

Cite

@article{arxiv.2006.14466,
  title  = {Splits with forbidden subgraphs},
  author = {Maria Axenovich and Ryan R. Martin},
  journal= {arXiv preprint arXiv:2006.14466},
  year   = {2025}
}

Comments

12 pages; In v1 and v2, there were missing floors and ceilings in Theorem 5 that in some cases would lead to an incorrect result. In v3, we correct this minor error

R2 v1 2026-06-23T16:37:36.613Z