Abstract:The NP-hard problems such as SAT (the propositional satisfiability problem) are long-term issues in the research of computer algorithms and theory. Some of the most successful and powerful methods for SAT problems, like GSAT, randomized hill-climbing, Tabu Search, and Simulated Annealing, can be cast into the framework of randomized algorithms. Randomized algorithms are playing a more and more important role in solving SAT problems because of their simplicity, speed and robustness. But the conventional randomized algorithms for SAT problems are mainly based on searching for satisfiable solutions. If no satisfiable solutions are found, it is difficult to justify the satisfiability of SAT problems correctly. In this paper, we propose a new idea that is to estimate the number of unsatisfiable solutions of a Conjunctive Normal Form (CNF) formula and to test the satisfiability, which is different from that of the conventional randomized algorithms. Even no satisfiable solutions can be found, it is still possible to test the satisfiability of the SAT problem. Based on the above idea, we design two randomized algorithms for SAT problems with time complexity of O(m 2), where m is the number of clauses in a CNF formula. For given accuracy parameter e and confidence parameter δ, they both give a mathematical expectation between 1-e and 1+e times the number of unsatisfiable solutions with probability at least 1-δ. If the expectation is less than 2 n, where n is the number of variables in a CNF formula, then the SAT instance is satisfiable, otherwise unsatisfiable. Furthermore, SatTest2 can estimate the number of unsatisfiable solutions while searches for a satisfiable solution, which is an important improvement of the conventional randomized algorithms. Experimental results, conducted on a number of SAT instances near the phase transition, indicate that our methods lead to significant performance improvements for hard SAT problems.

Solving SAT problem with a revised hitting set algorithm

An Efficient Resolution Based Algorithm for SAT

Improving probability selection based weights for satisfiability problems

Complete SAT Solver Based on Set Theory.

Solving SAT Problem by Heuristic Polarity Decision-Making Algorithm

Solving SAT by Algorithm Transform of Wu’s Method

Solving SAT Problems by Orthogonal Method

An Efficient Branch-and-Bound Solver for Hitting Set

EXPLORATION OF HARD TO SOLVE 3-SAT PROBLEMS

O(m~2) Time Randomized Algorithms for SAT Problem

A Heuristic Complete Algorithm for SAT Problem by Approximation Solution

Complete Boolean Satisfiability Solving Algorithms Based on Local Search

A survey of SAT solver

General Method for Solving Four Types of SAT Problems

An algorithm based on tabu search for satisfiability problem

Enhance SAT Conflict Analysis for Model Checking

A Hopeful CNF-SAT Algorithm - Its High Efficiency, Industrial Application and Limitation.

Between SAT and UNSAT: The Fundamental Difference in CDCL SAT

Clustering and Partition Based Divide and Conquer for SAT Solving

SAT solver with static and dynamic ordering decision making

AutoSAT: Automatically Optimize SAT Solvers via Large Language Models