Mittagsseminar (in cooperation with A. Steger, D. Steurer and B. Sudakov)

Date and Time: Tuesday, October 02, 2007, 12:15 pm

Duration: This information is not available in the database

Location: OAT S15/S16/S17

Speaker: Michael Krivelevich (Tel Aviv Univ.)

Avoiding small subgraphs in Achlioptas processes

Consider the following model of random graphs. We are given a monotone graph property P (Hamiltonicity, non-3-colorability, containment of a copy of a fixed graph H etc.) and an integer parameter r>=1. At each round, we are presented with r random edges from the set of edges of the complete graph K_n on n vertices and are asked to choose one of them. The task is to design an online edge choice algorithm that will be able to postpone (avoid) or to facilitate (embrace) almost surely the appearance of P. This model is sometimes called the Achlioptas process, after Dimitris Achlioptas who suggested it for the case r=2 and the property P being the existence of a linear sized connected component (the so called giant component) in the graph. The latter setting is essentially the only one that has been studied extensively so far.

In the work to be presented, we were able to achieve a satisfactory solution for the case where the task is to avoid the appearance of a fixed graph H, and H is either a cycle, or a complete graph, or a complete bipartite graph. The answer is quite surprising and depends heavily on the parameter r. For example, the threshold for avoiding K_4 in the Achlioptas process with parameter r=2 is n^{28/19}.

Joint work with Po-Shen Loh and Benny Sudakov, Princeton University.

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

Previous talks by year:   2024  2023  2022  2021  2020  2019  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