Monotone Increasing Properties and Their Phase Transitions in Uniform Random Intersection Graphs

Uniform random intersection graphs have received much interest and been used in diverse applications. A uniform random intersection graph with $n$ nodes is constructed as follows: each node selects a set of $Kn$ different items uniformly at random from the same pool of $Pn$ distinct items, and two nodes establish an undirected edge in between if and only if they share at least one item. For such graph denoted by $G(n, Kn, Pn)$, we present the following results in this paper. First, we provide an exact analysis on the probabilities of $G(n, Kn, Pn)$ having a perfect matching and having a Hamilton cycle respectively, under $Pn = \omega\big(n (\ln n)^5\big)$ (all asymptotic notation are understood with $n \to \infty$). The analysis reveals that just like ($k$-)connectivity shown in prior work, for both properties of perfect matching containment and Hamilton cycle containment, $G(n, Kn, Pn)$ also exhibits phase transitions: for each property above, as $Kn$ increases, the limit of the probability that $G(n, Kn, Pn)$ has the property increases from $0$ to $1$. Second, we compute the phase transition widths of $G(n, Kn, Pn)$ for $k$-connectivity (KC), perfect matching containment (PMC), and Hamilton cycle containment (HCC), respectively. For a graph property $R$ and a positive constant $a < \frac{1}{2}$, with the phase transition width $dn(R, a)$ defined as the difference between the minimal $Kn$ ensuring $G(n, Kn, Pn)$ having property $R$ with probability at least $1-a$ or $a$, we show for any positive constants $a<\frac{1}{2}$ and $k$: (i) If $Pn=\Omega(n)$ and $Pn=o(n\ln n)$, then $dn(KC, a)$ is either $0$ or $1$ for each $n$ sufficiently large. (ii) If $Pn=\Theta(n\ln n)$, then $dn(KC, a)=\Theta(1)$. (iii) If $Pn=\omega(n\ln n)$, then $dn(KC, a)=\omega(1)$. (iv) If $Pn=\omega\big(n (\ln n)^5\big)$, $dn(PMC, a)$ and $dn(HCC, a)$ are both $\omega(1)$.