PB'06 evaluation: available results
WARNING!
This site gives some information based on the answers provided by the different solvers. Not all answers given by a solver can be checked efficiently. Therefore, some answers given by a solver may be wrong but this may be unnoticed by the evaluation environment.
Here are the checks made for each answer.
- SATISFIABLE answers:
The model provided by the solver is given to a verifier program which checks that all constraints are actually satisfied. These answers are the only ones that are guaranteed to be true.
- UNSATISFIABLE answers:
The only check that is made is to verify that no other solver found a solution to this instance. An UNSATISFIABLE answer is not reliable. Some solvers may wrongly indicate UNSATISFIABLE but unless another solver finds a model, this will remain undetected.
- OPTIMUM FOUND answers:
Beyond verifying the model and the value of the objective function, the only check that is made is to verify that no other solver found a better solution to this instance. An OPTIMUM FOUND answer is not reliable. Some solvers may wrongly indicate OPTIMUM FOUND but unless another solver finds a better model, this will remain undetected.
General information
Global results of solvers
Individual results of solvers
Results for each bench
- Description of categories and link to benchmarks download
- Results for category OPT-BIGINT (optimization, big integers, 355 instances)
- Results for category OPT-MEDINT (optimization, medium integers, 25 instances)
- Results for category OPT-SMALLINT (optimization, small integers, 954 instances)
- Results for category SATUNSAT-BIGINT (no optimization, big integers, 2 instances)
- Results for category SATUNSAT-MEDINT (no optimization, medium integers, 12 instances)
- Results for category SATUNSAT-SMALLINT (no optimization, small integers, 405 instances)
- Results for all categories
Some graphs