A Learned Clause Deletion Strategy Based on Distance Ratio

Abstract

learnt clauses are an important component part of the CDCL-SAT solver, every conflict can produce learnt clauses, but sometimes the number of clause increases explosively. In order to avoid taking up too much memory, and speeding up the solving, it is necessary to evaluate the value of the clause and delete learnt clauses which are utilized low. Based on this, this paper puts forward to a kind of learnt clauses deletion strategy based on distance ratio. The idea of this strategy is to balance the value of Literal Blocks Distance (LBD), put forward the concept of distance ratio, sort clauses according to the distance ratio, and combine clause activity to form a more efficient deletion strategy. SATLIB basic examples and SAT competition instances are used to test, and the comparative analysis results show that new method can reduce the solving time, and hence improve solution efficiency.