The length of random subsets of Boolean lattices (Q4948020)

Among various constructions of random posets, the random poset \(P(n,p)\) of all subsets of a (standard) \(n\)-set \((\{1,\dots, n\})\), each drawn with probability \(p\) and ordered by set-inclusion, for \(p= 1\) yields the \(n\)-cube (Boolean algebra, lattice etc.) and for \(p< 1\) a subposet of that structure. Being random, \(P(n,p)\) varies from draw-to-draw and hence numerical invariants (e.g., height or length and width) become random variables whose statistical behavior describing certain ``approximate'' properties is then of interest and for which asymptotic properties \((n\to\infty)\) are also meaningful. In discussing these asymptotic properties one may also vary \(p\) (e.g., by requiring \(p\geq e/n\)). Techniques used to develop results often depend on the model used and in this case the arguments are ``random-graphian'' in nature, unsurprisingly perhaps by these authors. As always required, the proofs are a delicate mix of tricky inequalities and good structural insight expertly put together to provide intriguing results, e.g., Corollary 3, Theorem 1, if \(c^n= o(p)\) for any \(0< c< 1\) and \(p= O(1/n)\), then a.s. length \[ P(n, p)= \Biggl(1+ O(1) {n\over \log(1/p)} \Biggl(\log n-\log\log{1\over p}\Biggr)\Biggr), \] and many others.