English

On Coloring Random Subgraphs of a Fixed Graph

Combinatorics 2018-05-03 v2 Data Structures and Algorithms Probability

Abstract

Given an arbitrary graph GG we study the chromatic number of a random subgraph G1/2G_{1/2} obtained from GG by removing each edge independently with probability 1/21/2. Studying χ(G1/2)\chi(G_{1/2}) has been suggested by Bukh~\cite{Bukh}, who asked whether E[χ(G1/2)]Ω(χ(G)/log(χ(G)))\mathbb{E}[\chi(G_{1/2})] \geq \Omega( \chi(G)/\log(\chi(G))) holds for all graphs GG. In this paper we show that for any graph GG with chromatic number k=χ(G)k = \chi(G) and for all dk1/3d \leq k^{1/3} it holds that Pr[χ(G1/2)d]<exp(Ω(k(kd3)d3))\Pr[\chi(G_{1/2}) \leq d] < \exp \left(- \Omega\left(\frac{k(k-d^3)}{d^3}\right)\right). In particular, Pr[G1/2 is bipartite]<exp(Ω(k2))\Pr[G_{1/2} \text{ is bipartite}] < \exp \left(- \Omega \left(k^2 \right)\right). The later bound is tight up to a constant in Ω()\Omega(\cdot), and is attained when GG is the complete graph on kk vertices. As a technical lemma, that may be of independent interest, we prove that if in \emph{any} d3d^3 coloring of the vertices of GG there are at least tt monochromatic edges, then Pr[χ(G1/2)d]<eΩ(t)\Pr[\chi(G_{1/2}) \leq d] < e^{- \Omega\left(t\right)}. We also prove that for any graph GG with chromatic number k=χ(G)k = \chi(G) and independence number α(G)O(n/k)\alpha(G) \leq O(n/k) it holds that E[χ(G1/2)]Ω(k/log(k))\mathbb{E}[\chi(G_{1/2})] \geq \Omega \left( k/\log(k) \right). This gives a positive answer to the question of Bukh for a large family of graphs.

Keywords

Cite

@article{arxiv.1612.04319,
  title  = {On Coloring Random Subgraphs of a Fixed Graph},
  author = {Igor Shinkar},
  journal= {arXiv preprint arXiv:1612.04319},
  year   = {2018}
}
R2 v1 2026-06-22T17:22:40.348Z