Clique factors in Kneser graphs

TL;DR

This paper investigates clique factors and Hamiltonian properties in Kneser graphs, providing new bounds and partial results towards understanding powers of Hamiltonian cycles and clique decompositions.

Contribution

It establishes a near-complete clique partition result for Kneser graphs with large n and extends Hamiltonicity results to broader parameter ranges.

Findings

For n ≥ ℓ^3 k, all but at most ℓ-1 vertices can be covered by cliques of size ℓ.

Extended Hamiltonicity for Kneser graphs to n ≥ 4k when k ≥ 4.

Provided partial progress towards linear bounds for clique factors in Kneser graphs.

Abstract

For $k,n\in \mathbb{N}$, the Kneser graph $K(n,k)$ is the graph with vertex set $V=[n]^{(k)}$ and edge set $E=\{\{x,y\} \in V^{(2)}: x\cap y=\emptyset\}$. Chen proved that for $n\geq 3k$, Kneser graphs are Hamiltonian. Similarly as for graphs with Hajnal's and Szemer\'edi's result about a minimum degree condition for clique factors and the P\'osa-Seymour Conjecture together with its solution for large graphs due to Koml\'os, S\'ark\"ozy, and Szemer\'edi, the next step is to ask for clique factors and powers of Hamiltonian cycles in Kneser graphs. For $k,\ell\in \mathbb{N}$, let $n(k,\ell)$ be the smallest integer such that for $n\geq n(k,\ell)$, $K(n,k)$ contains the $\ell$-th power of a Hamiltonian cycle. Katona conjectured that for all but finitely many exceptions, $n(k,\ell)=(\ell+1)k+1$ holds. In particular, it would be interesting to know whether $n(k,\ell)$ is linear in $k$ (for fixed $\ell$). So far this is not known for $k\geq 2$. In this note, we take a first step towards such a linear bound by proving that for $\ell\in \mathbb{N}$, $k\geq \ell$ and $n\geq \ell ^3k$, all but at most $\ell-1$ vertices of $K(n,k)$ can be partitioned into cliques of size $\ell$. Further, we use our methods to extend a short proof due to Chen and F\"uredi that $K(n,k)$ is Hamiltonian for $n\geq 3k$ and $k \mid n$ to all $n\geq 4k$ if $k\geq 4$.