Understanding the Empirical Hardness of NP-Complete Problems I

preview_player
Показать описание
Kevin Leyton-Brown, University of British Columbia
Algorithms and Uncertainty Boot Camp
Рекомендации по теме
Комментарии
Автор

Seems quite obvious why there would be a large variance for a DPLL type of solver near the center. There is a high degree of sequence dependence on how fast a satisfiable instance is found (or none found at all) near the phase transition. On the other hand, for the unsat by overdetermination instances, there is no sequence dependence of the time at all.

John-lfxf