Department of Computer Science | Institute of Theoretical Computer Science | CADMO

Prof. Emo Welzl and Prof. Bernd Gärtner

Mittagsseminar Talk Information |

**Date and Time**: Thursday, September 07, 2006, 12:15 pm

**Duration**: This information is not available in the database

**Location**: CAB G51

**Speaker**: Gregory B. Sorkin (IBM T.J. Watson Research Center)

The class Max 2-CSP includes Max Cut, Max 2-Sat, and weighted versions of more general problems. Most exact (exponential-time) algorithms have been targeted to specific problems in the class, for example solving Max Cut in time Otilde(2^{m/4}), where m is the number of edges (constraints, generally). I will show a simple algorithm which solves any Max 2-CSP in time Otilde(2^{19m/100}), making it the fastest known algorithm even for most Max 2-CSP special cases (with exceptions such as Maximum Independent Set). For random constraint graphs up to the giant-component threshold, the algorithm runs in expected linear time, an interesting result in the context of the experimentally observed phase transition in the hardness of solving random 3-Sat instances.

The talk will focus on a new problem class, Generalized Max 2-CSP, whose partition functions may be calculated by essentially the same algorithm but whose interest transcends the algorithm. It enables us, in the same time bound Otilde(2^{19m/100}), to solve counting problems like #Max 2-Sat, to sample maximizing solutions uniformly at random (or to sample over all solutions from other distributions, such as the Gibbs distribution), and to solve problems like Max Bisection and Max Clique which do not quite fit the Max 2-CSP paradigm.

Joint work with Alexander Scott (Oxford).

Upcoming talks | All previous talks | Talks by speaker | Upcoming talks in iCal format (beta version!)

Previous talks by year: 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996

Information for students and suggested topics for student talks

Automatic MiSe System Software Version 1.4803M | admin login