Christophe Lecoutre, Lakhdar Sais, Sébastien Tabary, Vincent Vidal
In this paper, nogood recording is investigated within the randomization and restart framework. Our goal is to avoid the same situations to occur from one run to the next one. More precisely, nogoods are recorded when the current cutoff value is reached, i.e. before restarting the search algorithm. Such a set of nogoods is extracted from the last branch of the current search tree. Interestingly, the number of nogoods recorded before each new run is bounded by the length of the last branch of the search tree. As a consequence, the total number of recorded nogoods is polynomial in the number of restarts. Experiments over a wide range of CSP instances demonstrate the effectiveness of our approach.
Subjects: 15. Problem Solving; 15.2 Constraint Satisfaction
Submitted: Oct 10, 2006