On the stochastic independence properties of hard-core distributions (Q1272181)

A probability measure \(p\) on the set \({\mathcal M}\) of matchings in a graph (or, more generally 2-bounded hypergraph) \(\Gamma\) is hard-core if for some \(\lambda :\Gamma\mapsto [0,\infty)\), the probability \(p(M)\) of \(M\in{\mathcal M}\) is proportional to \(\prod _{A\in M}\lambda (A)\). It is shown in this paper that such distributions have substantial approximate stochastic independence properties, e.g. the main theorem is: For every \(\delta \in (0,1)\), \(\eta >0\) and \(\ell\geq 1\), there exist \(\varepsilon >0\) and an integer \(D\) such that the following is true. Let \(\Gamma\) be a multigraph and \({\mathcal M}\) the set of matchings in \(\Gamma\). Let \(p\) be a hard-core distribution on \({\mathcal M}\) with marginals \(f_p\in (1-\delta)\text{MP}(\Gamma)\) (here MP\((\Gamma)\) is the matching polytope of \(\Gamma\)), and let \(\Gamma ' =\{A\in\Gamma :p_A>\varepsilon\}\). If \(F_1,\dots F_\ell\in\Gamma\) are pairwise at distance at least \(D\) in \(\Gamma '\), then Pr\((F_1,\dots F_\ell\in M)=\eta\prod_{i=1}^\ell p_{F_i}\). This result is obtained through a sequence of lemmata which are interesting at their own, too. The obtained results are very useful in the resolution of several questions involving the asymptotic behaviour of graphs and hypergraphs.