Large induced subgraphs of random graphs with given degree sequences

Angus Southwell, Nicholas Wormald

Published 2023-03-15Version 1

We study a random graph $G$ with given degree sequence $\boldsymbol{d}$, with the aim of characterising the degree sequence of the subgraph induced on a given set $S$ of vertices. For suitable $\boldsymbol{d}$ and $S$, we show that the degree sequence of the subgraph induced on $S$ is essentially concentrated around a sequence that we can deterministically describe in terms of $\boldsymbol{d}$ and $S$. We then give an application of this result, determining a threshold for when this induced subgraph contains a giant component. We also apply a similar analysis to the case where $S$ is chosen by randomly sampling vertices with some probability $p$, i.e. site percolation, and determine a threshold for the existence of a giant component in this model. We consider the case where the density of the subgraph is either constant or slowly going to $0$ as $n$ goes to infinity, and the degree sequence $\boldsymbol{d}$ of the whole graph satisfies a certain maximum degree condition. Analogously, in the percolation model we consider the cases where either $p$ is a constant or where $p \to 0$ slowly. This is similar to work of Fountoulakis in 2007 and Janson in 2009, but we work directly in the random graph model to avoid the limitations of the configuration model that they used.