Actin - Technical Report

TL;DR

This paper introduces Actin, a SAT solver that uses genetic programming to optimize initial variable activities based on CNF analysis, improving conflict reduction and solving time.

Contribution

It presents a novel approach of evolving initial variable activities for SAT solvers using genetic programming, enhancing performance over traditional zero-initialization.

Findings

Optimized initial activities reduce total conflicts.

Improved initialization counters CNF reordering effects.

Actin outperformed baseline solvers in SAT-Race 2006.

Abstract

The Boolean satisfiability problem (SAT) can be solved efficiently with variants of the DPLL algorithm. For industrial SAT problems, DPLL with conflict analysis dependent dynamic decision heuristics has proved to be particularly efficient, e.g. in Chaff. In this work, algorithms that initialize the variable activity values in the solver MiniSAT v1.14 by analyzing the CNF are evolved using genetic programming (GP), with the goal to reduce the total number of conflicts of the search and the solving time. The effect of using initial activities other than zero is examined by initializing with random numbers. The possibility of countering the detrimental effects of reordering the CNF with improved initialization is investigated. The best result found (with validation testing on further problems) was used in the solver Actin, which was submitted to SAT-Race 2006.