Mittagsseminar (in cooperation with J. Lengler, A. Steger, and D. Steurer)

Date and Time: Tuesday, October 03, 2006, 12:15 pm

Duration: This information is not available in the database

Location: OAT S15/S16/S17

Speaker: Amin Coja-Oghlan (HU Berlin)

Graph Partitioning via Adaptive Spectral Techniques

We study random instances of a general graph partitioning problem: the vertex set of the random input graph G consists of k classes V_1, . . . , V_k, and V_i-V_j-edges are present with probabilities p_{ij} independently. The main result is that with high probability a partition S_1, . . . , S_k of G that coincides with V_1, . . . , V_k on a huge subgraph core(G) can be computed in polynomial time via spectral techniques. The result covers the case of sparse graphs (average degree O(1)) as well as the massive case (average degree #V(G) - O(1)). Furthermore, the spectral algorithm is adaptive in the sense that it does not require any information about the desired partition beyond the number k of classes.

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

Previous talks by year:   2025  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