Notes on Generating Satisfiable SAT instances using random subgraph isomorphism

Abstract

We present a preprocessing procedure for the Subgraph Isomorphism problem, and report empirical results of applying it to Generating Satisfiable SAT instances using the Random Subgraph Isomorphism model. The experimental results show that the preprocessor makes the easy-hard-easy pattern of empirical hardness clear for all SAT solvers tested. Moreover, the use of the procedure does not significantly influence the location or the empirical hardness of the instances at the hardness peak, thus preserving the model's main characteristics. © 2010 Springer-Verlag Berlin Heidelberg.