BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:From linear programming to statistics: Fast algorithms for samplin
 g based on interior point methods - Martin Wainwright (UC Berkeley)
DTSTART:20171124T153000Z
DTEND:20171124T163000Z
UID:TALK96199@talks.cam.ac.uk
CONTACT:Quentin Berthet
DESCRIPTION:Sampling from distributions is a core challenge in\nstatistics
 \, computer science and operations research.  An evolving\nbody of work is
  showing how algorithms from optimization can be\nmodified so as to sample
  from distributions.  In this talk\, we describe and analyze some novel al
 gorithms\, based on modifications of\ninterior point methods used in linea
 r programming\, for sampling points\nuniformly from polytopes.  Such sampl
 ing algorithms are useful for\nvolume computation\, contigency table analy
 sis\, post selection\ninference\, and the hard disk problem in statistical
  physics\, among\nother applications.  We propose and analyze the mixing t
 imes of two new Markov chain methods\, referred as the Vaidya and John wal
 ks\, both of which\nyield substantial improvements over the state-of-the-a
 rt Dikin walk.\n\nBased on joint work with:  Yuansi Chen\, Raaz Dwivedi\, 
 and Bin Yu\nPre-print:  https://arxiv.org/abs/1710.08165
LOCATION:MR12
END:VEVENT
END:VCALENDAR