This collection of benchmark problems is still under development.  Some data sets are missing and some data sets will need to have their format changed (to ensure uniformity). The next revision will be available early in September.

A necessary element for the development of fast software systems is a comprehensive suite of benchmark problems. Answer-set programming (ASP) has been developing rapidly in the recent years. Several implementations were proposed. Some of them have already been quite thoroughly tested. Others have not. It is clear that further progress in answer-et programming depends to a large degree on establishing a collection of benchmarks on which ASP implementations should be tested and with respect to which they will be compared.  This sentiment was voiced several times in the past, most recently at the AAAI Spring 2001 Symposium on Answer-Set Programming.

The following  suite of benchmarks was developed for the special session on systems and implementations at Logic Programming and Nonmonotonic Reasoning Conference LPNMR'01 in Vienna.

To facilitate solving these problems in your own ASP formalism, we provide below a number of programs in smodels and dlv that were written to solve them. More programs will be available in the near future.

15-puzzle (smodels)
n-queens (smodels)
Sokoban (dlv)
Schur numbers (smodels)
Hamilton-cycle program (smodels)
Vertex-cover (smodels)
2QBF(AE) (dlv)

Proposed  by Mirek Truszczynski, Nicola Leone and Ilkka Niemela with additional suggestions and contributions from Vladimir Lifschitz and Deborah East. Send your comments to