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

Theory of Combinatorial Algorithms

Prof. Emo Welzl and Prof. Bernd Gärtner

Mittagsseminar (in cooperation with M. Ghaffari, A. Steger and B. Sudakov)

Mittagsseminar Talk Information

Date and Time: Thursday, October 25, 2012, 12:15 pm

Duration: 30 minutes

Location: CAB G51

Speaker: Sebastian Millius

Towards a Generalization of the PPSZ Algorithm to Larger Domains, Part II

Since Timon Hertli's improved analysis (you recall his mittagsseminar talk of May 19, 2011), the PPSZ algorithm due to Paturi, Pudlák, Saks and Zane is the fastest known randomized algorithm for the k-SAT problem. In joint work with Hertli, Moser, Scheder and Szedlák (Bachelors's thesis), we demonstrate that the PPSZ approach can be generalized to the case of (d,k)-ClSP which is the variant of k-SAT where the variables can take any number d of values, rather than just two. For many cases of d and k, we thereby obtain the fastest known (d,k)-ClSP algorithm.

Last week, Robin Moser suggested a generalization of the PPSZ algorithm to (d,k)-ClSP and sketched an analysis in the simpler case where we assume the ClSP to have exactly one satisfying assignment. For this case, we obtain the fastest known algorithm for all d and k. This was largely based on the Bachelor's thesis by May Szedlák.

In this follow-up talk, I will explain the result of my Master's thesis where we establish that for many important cases of d and k, the aforementioned analysis can be generalized so as to hold without the uniqueness assumption, by means of similar proof techniques as provided by Hertli.


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