The invariable Ewens distribution

Series
Stochastics Seminar
Time
Thursday, September 8, 2016 - 3:05pm for 1 hour (actually 50 minutes)
Location
Skiles 006
Speaker
Matthew Junge – Duke University – jungem@math.duke.eduhttp://www.mathjunge.com
Organizer
Michael Damron
Form a multiset by including Poisson(1/k) copies of each positive integer k, and consider the sumset---the set of all finite sums from the Poisson multiset. It was shown recently that four such (independent) sumsets have a finite intersection, while three have infinitely many common elements. Uncoincidentally, four uniformly random permutations will invariably generate S_n with asymptotically positive probability, while three will not. What is so special about four? Not much. We show that this result is a special case of the "ubiqituous" Ewens sampling formula. By varying the distribution's parameter we can vary the number of random permutations needed to invariably generate S_n, and, relatedly, the number of Poisson sumsets to have finite intersection. *Joint with Gerandy Brita Montes de Oca, Christopher Fowler, and Avi Levy.