English

Hitting times for Shamir's Problem

Combinatorics 2020-08-05 v1

Abstract

For fixed r3r\geq 3 and nn divisible by rr, let H=Hn,Mr{\mathcal H}={\mathcal H}^r_{n,M} be the random MM-edge rr-graph on V={1,,n}V=\{1,\ldots ,n\}; that is, H{\mathcal H} is chosen uniformly from the MM-subsets of K:=(Vr){\mathcal K}:={V \choose r} (:= \{\mbox{rsubsetsof-subsets of V}\}). Shamir's Problem (circa 1980) asks, roughly, for what M=M(n)M=M(n) is H{\mathcal H} likely to contain a perfect matching (that is, n/rn/r disjoint rr-sets)? In 2008 Johansson, Vu and the author showed that this is true for M>CrnlognM>C_rn\log n. More recently the author proved the asymptotically correct version of that result: for fixed C>1/rC> 1/r and M>CnlognM> Cn\log n, P({\mathcal H} ~\mbox{contains a perfect matching})\rightarrow 1 \,\,\, \mbox{as n\rightarrow\infty}. The present work completes a proof, begun in that recent paper, of the definitive "hitting time" statement: \mboxTheorem.\mbox{Theorem.} If A1, A_1, \ldots ~ is a uniform permutation of K{\mathcal K}, Ht={A1At}{\mathcal H}_t=\{A_1\dots A_t\}, and T=min{t:A1At=V}, T=\min\{t:A_1\cup \cdots\cup A_t=V\}, then P({\mathcal H}_T ~\mbox{contains a perfect matching})\rightarrow 1 \,\,\, \mbox{as n\rightarrow\infty}.

Cite

@article{arxiv.2008.01605,
  title  = {Hitting times for Shamir's Problem},
  author = {Jeff Kahn},
  journal= {arXiv preprint arXiv:2008.01605},
  year   = {2020}
}

Comments

40 pages. arXiv admin note: text overlap with arXiv:1909.06834

R2 v1 2026-06-23T17:38:09.255Z