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

Prof. Emo Welzl and Prof. Bernd Gärtner

Mittagsseminar Talk Information |

**Date and Time**: Tuesday, October 13, 2015, 12:15 pm

**Duration**: 30 minutes

**Location**: CAB G51

**Speaker**: Christoph Koch (TU Graz)

Bootstrap percolation with activation parameter $r>1$ is a deterministic "infection" process on a hypergraph evolving in rounds: Initially some subset of vertices is infected. Subsequently in each round the infection spreads to all uninfected vertices with at least $r$ infected neighbours. Once a vertex becomes infected, it remains so forever.

In this talk we consider bootstrap percolation on the binomial random $k$-uniform hypergraph $H^k(n,p)$, whose vertex set is $\{1,\dots,n\}$ and where each edge consists of precisely $k$ distinct vertices and is present with probability $p$ independently. The main question is whether the infection can spread from a small set of initially infected vertices to almost all vertices. We determine the exact threshold for this property with respect to the number of initially infected vertices in the regime $n^{-1}\ll p\ll n^{-1/r}$, i.e. a threshold function $a_c=a_c(r,n,p)$ such that for any constant $\varepsilon>0$ with probability tending to $1$ as $n\to\infty$ the following two properties hold: Infecting at most $(1-\varepsilon)a_c$ vertices initially leads to no significant spread of the infection; While an initial infection of at least $(1+\varepsilon)a_c$ vertices suffices to spread the infection to almost all vertices.

This extends a result by Janson, T. Luczak, Turova and Vallier concerning the graph case $k=2$.

Joint work with Mihyun Kang and Tamas Makai.

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