Random perfect graphs

From MaRDI portal



Abstract: We investigate the asymptotic structure of a random perfect graph Pn sampled uniformly from the perfect graphs on vertex set 1,ldots,n. Our approach is based on the result of Pr"omel and Steger that almost all perfect graphs are generalised split graphs, together with a method to generate such graphs almost uniformly. We show that the distribution of the maximum of the stability number alpha(Pn) and clique number omega(Pn) is close to a concentrated distribution L(n) which plays an important role in our generation method. We also prove that the probability that Pn contains any given graph H as an induced subgraph is asymptotically 0 or frac12 or 1. Further we show that almost all perfect graphs are 2-clique-colourable, improving a result of Bacs'o et al from 2004; they are almost all Hamiltonian; they almost all have connectivity kappa(Pn) equal to their minimum degree; they are almost all in class one (edge-colourable using Delta colours, where Delta is the maximum degree); and a sequence of independently and uniformly sampled perfect graphs of increasing size converges almost surely to the graphon WP(x,y)=frac12(mathbb1[xle1/2]+mathbb1[yle1/2]).











This page was built for publication: Random perfect graphs

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4625032)