Summer 2004 Surprise Seminars!We invite all faculty, researchers, and students to join us for our summer seminars (sometimes called Surprise Seminars).
Thursday, August 19, 12:00, Mason room 107.I believe Sekhar will be talking about spin glasses?
Thursday, August 5Jay will give a brief warm-up talk for the Joint Meetings.
Thursday, July 29Sekhar will first present a concentration result useful for proving the self-averaging property of spin glasses. The proof of this result is based on the "smart path method." Then he will introduce the notion of overlaps and the role they play in spin glasses.
Thursday, July 22.Sekhar will first present an alternative proof of Talagrand's convex distance exponential bound based on the "entropy method" introduced by Ledoux and Massart. Then we will present another concentration result useful for proving the self-averaging property of spin glasses. The proof of this result is based on the "smart path method." Finally, if there is time, he will introduce the notion of overlaps and the role they play in spin glasses.
Thursday, July 15.Stephan Winkler will motivate (using a bin packing problem) and prove Talagrand's "convex hull concentration inequality", which improves on the bound derived last week.
Thursday, July 8.David will be speaking about some of the motivation for one of Talagrands concentration inequalities (multi-point control).
Thursday, July 1. John Hartigan, "How to win the Tour De France."The Tour runs from July 3 through July 25, starting with the Prologue in Liege, Belgium, and finishing on the Champs-Elysees in Paris. Really, this will happen on this date! Wait, didn't I say that last week? John's talk outline and the data set is available here. Post race question: Was Basso on John's list? But we'll forgive him for not anticipating Hamilton's dropping out with an injury, of course.
Thursday, June 24.David Pollard, concentration inequalities.
Thursday, June 17.David Pollard's summer seminar on concentration of measure and statistical mechanics will continue, with more by Sekhar Tatikonda: Free Energy: What is it Good For? - Part 3We will continue our review of Gibbs measures and disordered systems. Some of the main disordered models, including the REM and SK, will be discussed. We will motivate the use of Mean-Field models and discuss some of the basic results. If time permits we will present a quick introduction to the cavity method and the replica method.
Thursday, June 10.David Pollard's summer seminar on concentration of measure and statistical mechanics will continue, with more by Sekhar Tatikonda: Free Energy: What is it Good For? - Part 2We will continue our review of Gibbs measures and disordered systems. Some of the main disordered models, including the REM and SK, will be discussed. We will motivate the use of Mean-Field models and discuss some of the basic results. If time permits we will present a quick introduction to the cavity method and the replica method.
Thursday, June 3.David Pollard's summer seminar on concentration of measure and statistical mechanics will kick off Thursday with a general overview by Sekhar Tatikonda:Physicists have developed many techniques for determining equilibrium properties of statistical mechanical systems. These techniques have also found use in combinatorial optimization problems. Some of these techniques include: the mean field approximation, the cavity method, and the replica method. These techniques have made remarkable predictions but at present there does not exist a rigorous mathematical basis for them. Talagrand and others have started on a program to make these techniques rigourous. These techniques are based on tools from convexity and concentration of measure. This summer we hope to understand these techniques as well.
Thursday, May 27.Jake introduced his project on real estate clustering.
Thursday, May 20.David was selected as the Surprise Seminar speaker, and posed some perl questions, with supporting documents posted.Thursday, May 13.Jay Emerson will consider a data set of Hamden and New Haven housing values in the Edgehill Road neighborhood (near the town line). The goal: to explore the use of R for 'dirty data parsing' in place of Perl. Jay will provide a brief (and by no means comprehensive) introduction to both Perl and R, using this data set as an example. This data set was used on the Practical Exam, December 2004. The data and associated files are available here.
|