Nearly tight bound for rainbow clique subdivisions in properly edge-colored graphs and applications
Peiru Kuang, Yan Wang
TL;DR
We study rainbow analogues of subdivision problems in edge-colored graphs, focusing on rainbow subdivisions of cliques. The authors develop a robust sublinear expander framework and a two-stage thinning–sprinkling method to force rainbow clique subdivisions with controlled path lengths, showing that a graph on $n$ vertices with average degree at least $C t^2 \log n (\log \log n)^6$ contains a rainbow $TK_t$ where each edge of $K_t$ is replaced by a path of length $O(\log n \cdot \log \log n)$. For fixed $t$, this bound is essentially tight, resolving a key question on rainbow clique subdivisions and yielding a rainbow-cycle consequence. The results have broad applications in additive combinatorics, number theory, and coding theory, including bounds on dissociated sets, $B_h[g]$-sequences in groups, and dimensions of locally correctable codes.
Abstract
An edge-colored graph is said to be rainbow if all its edges have distinct colors. In this paper, we study the rainbow analogue of a fundamental result of Mader [\emph{Math. Ann.} \textbf{174} (1967), 265--268] on the existence of subdivisions in graphs with large average degree. This is part of the study of rainbow analogues of classical Turán problems, a framework systematically introduced by Keevash, Mubayi, Sudakov and Verstraëte [\emph{Combin. Probab. Comput.} \textbf{16} (2007), 109--126]. We prove that every properly edge-colored graph on $n$ vertices with average degree at least $t^2(\log n)^{1+o(1)}$ contains a rainbow subdivision of $K_t$. When $t$ is a constant, this bound is tight up to the $o(1)$ term. So it essentially resolves a question raised by Jiang, Methuku and Yepremyan [\emph{European J. Combin.} \textbf{110} (2023), 103675] on rainbow clique subdivisions, and also implies a result of Alon, Bucić, Sauermann, Zakharov and Zamir [\emph{Proc. Lond. Math. Soc.} \textbf{130} (2025), e70044] on rainbow cycles. In addition, we present several applications of our result to problems in additive combinatorics, number theory and coding theory.
