Paperlane

no. 17 · Wednesday 18 November 2026

On Threshold Probabilities for the Stability of the Independence Number of a Random Subgraph

Vsevolod A. Pokhachevskiy and Andrei Raigorodskii

math.CO · 18 pp

Abstract

For an arbitrary sequence of graphs $G_n$, we consider a random spanning subgraph $G_{n,p}$ in which each edge of the graph $G_n$ survives independently of the others with probability $p$. We obtain a general upper bound for the independence number $α(G_{n,p})$, from which, for an arbitrary probability $p=p(n)$, we derive sufficient conditions for asymptotic stability, i.e., for the equality $α(G_{n,p})=(1+o(1))α(G_n)$ asymptotically almost surely, as well as upper bounds for the independence number of the random subgraph, including unconditional ones; for $p=1/2$ they recover A. M. Raigorodskii's results of 2017 and 2026. Under additional conditions on the structure of the independent sets of the graph $G_n$, we find, up to a constant factor, the threshold probabilities for both asymptotic and exact stability (the equality $α(G_{n,p})=α(G_n)$). Special cases include results on the graphs $G(n,r,<s)$ for growing $r$ and $s$, including Kneser graphs for growing $r$; we also give examples of complete multipartite graphs, graphs on the blocks of Steiner systems, and $q$-analogues of Kneser graphs.

comments
18 pages
fields
Combinatorics, Probability
pdf ↗ page arXiv:2611.22185 ↗