Algorithms for Streaming Graphs

An algorithm solving a graph problem is usually expected to have fast random access to the input graph G and a working memory being able to store G completely. These powerful assumptions are put in question by massive graphs that exceed common working memories and that can only be stored on disks or even tapes. Here, random access is very time-consuming. To tackle massive graphs stored on external memories, the semi-streaming model has been proposed. It permits a working memory of restricted size and forbids random access to G. In contrast, the input is assumed to be a stream of edges in arbitrary order. In this book we develop algorithms in the semi-streaming model approaching different graph problems. For the problems of testing graph connectivity and bipartiteness and for the computation of a minimum spanning tree, we show how to obtain optimal running times. For the intractable problem of finding a maximum weighted matching, we present the best known approximation algorithm. Finally, we show the minimum and the maximum cut problem in a graph both to be intractable in the semi-streaming model and give algorithms that approximate respective solutions in a randomized fashion.

Autorentext

The author holds a diploma in computer science and received hisPhD in the same field at the Humboldt-University Berlin in 2009.From 2006 to 2009 he worked as a researcher at the DFG researchcenter Matheon at Berlin. There he was concerned with designingand analyzing data stream algorithms for graphs modeling complexnetworks.