On the chromatic number of random geometric graphs

TL;DR

This paper investigates the asymptotic behavior of the chromatic number in random geometric graphs, revealing a phase transition in the ratio of chromatic to clique numbers around a critical threshold.

Contribution

It extends previous work by precisely characterizing the phase change in the chromatic number for the critical range of connection radius.

Findings

Determines the limit of the scaled chromatic number as n grows large.

Identifies a sharp threshold t_0 for the ratio of chromatic to clique number.

Shows the ratio tends to 1 below t_0 and exceeds 1 above t_0.

Abstract

Given independent random points $X_1,...,X_n\in\eR^d$ with common probability distribution $\nu$, and a positive distance $r=r(n)>0$, we construct a random geometric graph $G_n$ with vertex set $\{1,...,n\}$ where distinct $i$ and $j$ are adjacent when $\norm{X_i-X_j}\leq r$. Here $\norm{.}$ may be any norm on $\eR^d$, and $\nu$ may be any probability distribution on $\eR^d$ with a bounded density function. We consider the chromatic number $\chi(G_n)$ of $G_n$ and its relation to the clique number $\omega(G_n)$ as $n \to \infty$. Both McDiarmid and Penrose considered the range of $r$ when $r \ll (\frac{\ln n}{n})^{1/d}$ and the range when $r \gg (\frac{\ln n}{n})^{1/d}$, and their results showed a dramatic difference between these two cases. Here we sharpen and extend the earlier results, and in particular we consider the `phase change' range when $r \sim (\frac{t\ln n}{n})^{1/d}$ with $t>0$ a fixed constant. Both McDiarmid and Penrose asked for the behaviour of the chromatic number in this range. We determine constants $c(t)$ such that $\frac{\chi(G_n)}{nr^d}\to c(t)$ almost surely. Further, we find a "sharp threshold" (except for less interesting choices of the norm when the unit ball tiles $d$-space): there is a constant $t_0>0$ such that if $t \leq t_0$ then $\frac{\chi(G_n)}{\omega(G_n)}$ tends to 1 almost surely, but if $t > t_0$ then $\frac{\chi(G_n)}{\omega(G_n)}$ tends to a limit $>1$ almost surely.