Backbone Guided Dynamic Local Search for Propositional Satisfiability

This comparative study examines the impact of backbone guided heuristics on the performance of dynamic local search methods. We study alternatives to the backbone membership estimation problem, discuss how our proposed estimation phase addresses it, and discuss how this information is integrated in the host methods. Backbone guidance results in significantly faster dynamic local search on the large problems tested, but it is of questionable use for small problem domains where the cost of backbone estimation alone represents a significant part of the total search cost.