PARTY - A Software Library for Graph Partitioning

The problem of partitioning a graph into a number of pieces is one of the fundamental tasks in computer science and has a number of applications e.g. in computational mechanics or VLSI design. Finding optimal partitions according to diierent measures is in most cases NP-complete. Nevertheless, a large number of eecient partitioning heuristics have been developed during recent years. The performance of these methods in terms of computation time as well as quality of approximation is heavily innuenced by choices of parameters and certain implementation details. Fortunately, the partitioning problem itself is clearly deened and its description leads to a small interface. Thus, eecient implementations of approximation heuristics can be re-used for diierent applications. The PARTY partitioning library serves a variety of different partitioning methods in a very simple and easy way. Instead of implementing the methods directly, the user may take advantage of the ready implemented methods of the library. All implementations include latest developments to increase the performance of the partitioning heuristics. Two kinds of interfaces allow the use as stand-alone tool as well as the inclusion into application codes. The data-structures used as interface to PARTY are simple and easy to generate. Several research projects currently use the PARTY partitioning library to solve the partitioning problem.