Development of Black-box search and heuristics for OscaR solver for the XCSP3 competition

Designing an efficient and robust black-box search is one of the major concerns in Constraint Programming. In the past, various search heuristics and procedures have been proposed in the literature to tackle this problem. However, up to this day there is no unique best search strategy that is able to solve all problems optimally. For this reason, competitions such as the XCSP3 competition are organized to asses the performance of constraint solvers. This thesis compares existing search strategies as well as combinations of them in order to improve the black-box search used by the OscaR solver on constraint optimization problems. Experiments show that the newly adopted search strategy is able to improve the overall performance of the OscaR solver on the fast COP tracks of the 2018 XCSP3 competition.