Algorithm Engineering for large data sets

  • Contact: R. Dementiev, P. Sanders
We develop library of algorithms and data structures Stxxl (stxxl.sourceforge.net) that can handle efficiently huge data sets that do not fit into main memory of a computer. The library implements interfaces of the Standard C++ Library. Our library has applications in many areas.
Stxxl is the basis of our fast implementation of an algorithm that constructs full-text indexes of very large text collections. In
the framework of DFG project "Algorithm Engineering for Large Graphs and Memory Hierarchies" we have applied Stxxl to process huge graphs. In the past year we have worked on an implementation of efficient BFS algorithm, which is the basic component of many other graph algorithms. Another graph application we have worked on is the graph coloring of very large graphs. Graph coloring a very important subprocedure in many applications such as solving sparse systems of linear equations, resource allocation, scheduling, the construction and testing of VLSI circuits.