Starting from the Mancoosi meeting held in Nice on the January 7th and 8th 2010, we routinely run a Mancoosi internal Solver Competition, with the goal of providing real-world data and benchmarks for the the official Mancoosi International Solver Competition, and to test the competition infrastructure.
We make explicitly available through this web page all the input data, as well as the results found by each of the participating solvers.
Two fixed optimization criterias are used in the current version of the competition: they are both lexicographic combinations of four simple integer valued utility functions of a solution, which we summarize below and we detaile the criterias here
PARANOID: we want to answer the user request, minimizing the number of packages removed in the solution, and also the packages changed by the solution;
TRENDY: we want to answer the user request, minimizing the number of packages removed in the solution, minimising the number of outdated packages in the solution, and minimizing the number of extra packages installed;
Solvers are classified as
According to these two optimization criteria the current results are summarized here
A primer for CUDF is available at Primer
The (long) specification of CUDF, if you have any doubts, is here: tr3.pdf