BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:D-optimal design of experiments - Radoslav Harman\, Comenius Unive
 rsity
DTSTART:20111124T150000Z
DTEND:20111124T160000Z
UID:TALK33839@talks.cam.ac.uk
CONTACT:Elena Yudovina
DESCRIPTION:Consider an experiment consisting of a series of linear regres
 sion trials. The question is: How should we choose the trials in order to 
 obtain "as much information as possible" from the experiment? The most cla
 ssic answer is to choose the trials such that the resulting confidence ell
 ipsoid for the unknown regression parameter has minimum achievable volume\
 , which leads to the so-called D-optimal design. In the talk we will defin
 e the basic notions of exact and approximate designs and prove the fundame
 ntal "equivalence theorem" for D-optimality. We will also give several con
 crete examples of D-optimal exact and approximate designs for various line
 ar regression models\, relating the design problems to areas as different 
 as the theory of Brownian motion and the number of spanning trees of graph
 s.
LOCATION:CMS\, MR9
END:VEVENT
END:VCALENDAR