BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:CQIF Seminar
SUMMARY:Convex separation from convex optimization for lar
 ge-scale problems - Steve Brierley\, University of
  Cambridge
DTSTART;TZID=Europe/London:20161110T141500
DTEND;TZID=Europe/London:20161110T151500
UID:TALK68782AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/68782
DESCRIPTION:I'll present a new scheme to prove separation betw
 een a point and an arbitrary convex set S via call
 s to an oracle able to perform linear optimization
 s over S. Compared to other methods\, it has almos
 t negligible memory requirements and the number of
  calls to the optimization oracle does not depend 
 on the dimensionality of the underlying space. We 
 study the speed of convergence of the scheme under
  different promises on the shape of the set S and/
 or the location of the point\, validating the accu
 racy of the theoretical bounds with numerical exam
 ples. \n\nI will then present some applications of
  the scheme in quantum information theory. The alg
 orithm out-performs existing linear programming me
 thods for certain large scale problems\, allowing 
 us to certify nonlocality in bipartite scenarios w
 ith upto 42 measurement settings. I'll show how to
  use the algorithm to upper bound the visibility o
 f two-qubit Werner states\, hence improving known 
 lower bounds on Grothendieck's constant KG(3). Sim
 ilarly\, we compute new upper bounds on the visibi
 lity of GHZ states and on the steerability limit o
 f Werner states for a fixed number of measurement 
 settings.
LOCATION:MR4\, Centre for Mathematical Sciences\, Wilberfor
 ce Road\, Cambridge
CONTACT:Steve Brierley
END:VEVENT
END:VCALENDAR