An Algorithm Library for Massive Data Sets

Summer 03

[Chaos] We are working on a a software library STXXL which allows processing of huge data sets that do not fit into RAM memory but only on hard disks. The starting point of this library is the C++ standard template library STL. We want to implement a large part of its functionality to work also for huge inputs.


A course in algorithms and data structures. A working knowledge of C++. You are encouraged to take the lecture "Algorithm Library Design" at the same time. The Fopra project can be used as the programming project in the lecture.

Further Information

Privatdozent Dr. Peter Sanders, Building 46, R.308, sanders@mpi-sb.mpg.de
Roman Dementiev, Building 46, R.322, email

Possible Projects

The following list gives very preliminary ideas for possible projects. You can work on these projects in groups of one to three people. The implementations can range from prototypical to production quality (well written, documented, tested, and tuned) depending on difficulty and number of people involved. There may also be possibilities to continue projects as master's projects. The order of the list roughly follows our current priorities for completing the library.